ImMy!IWMIIIQQIQILHWZIIMIIll
7/

LIBRART

Michigan State
Unlverslty

—.—_

 

 

 

This is to certify that the

thesis entitled

presented by

Lyle Jay Buss

has been accepted towards fulfillment
of the requirements for

__M-3- fidegreeianLQgEntom0 Y

MS U i: an Waive Action/Equal Opportunity Institution

 

Date__ 8/7/97

0-7639

 

 

’

PLACER RETUQI‘I BOX to remove thle checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

l ‘“

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EVALUATION OF THREE EGG MASS SURVEY METHODS AND TWO
BIOLOGICAL CONTROL AGENTS FOR GYPSY MOTH MANAGEMENT

By

Lyle Jay Buss

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Entomology

1997

ABSTRACT

EVALUATION OF THREE EGG MASS SURVEY METHODS AND TWO
BIOLOGICAL CONTROL AGENTS FOR GYPSY MOTH MANAGEMENT

By

Lyle Jay Buss

Three gypsy moth (Lymantria dispar L.) egg mass survey methods were evaluated in
northwestern Lower Michigan. Although counts from the fixed-radius plot and lOO-tree
plot methods were better associated with subsequent defoliation, the timed walks required
much less time to complete. However, when common egg mass density thresholds were
used with egg mass counts to evaluate the study stands for hypothetical suppression, the
lOO-tree plot method provided fewer overspraying errors than the other 2 methods.

A Sunfleck Ceptometer® was used to measure leaf area index in hardwood stands at
the Michigan State University Kellogg Experimental Forest. The ceptometer provided
relatively consistent estimates, indicating that the ceptometer could be used to quantify
insect defoliation.

Two biocontrol agents, C otesia melanoscela (Ratzeburg) and Entomophaga
maimaiga Humber, Shimazu and Soper, were released at Kellogg Forest in 1994 and
1995. Their impact on the gypsy moth population was very low, possibly due to low

gypsy moth densities and other factors.

ACKNOWLEDGMENTS

First and foremost, I wish to thank my major professor, Dr. Deborah McCullough, for
her constant support. Her advice and encouragement have been invaluable to me over the
past several years, both as an advisor and as a friend. I am proud to have worked with
her. Advice contributed by my committee of Drs. Carl Ramm, Doug Landis, Bob Haack,
and Dave Smitley has been extremely helpful throughout this project. I am also indebted
to the crew at the W. K. Kellogg Forest (Greg Kowalewski, Karen Bushouse, Jim Curtis,
John Vigneron, and Don Alkema) and to several students during the summers (Kevin
Ellwood, Katie Albers, Liz Weber, Eric Black, Phil Garvin, and Nate Siegert). Suzanne
Thiem (Michigan State University) and Debbie Miller (USDA Forest Service) provided
valuable assistance in diagnosing insect pathogens.

The Forestry Division of the Michigan Department of Natural Resources provided
assistance and funding for the biocontrol releases. Appropriate permits for shipping and
release of the biocontrol agents were provided by the Michigan Department of Agriculture
and the U. S. Department of Agriculture, Animal and Plant Health Inspection Service,
Plant Protection and Quarantine (USDA APHIS PPQ). Testing of Entomophaga
maimaiga release soil for plant parasitic nematodes was conducted by the Nematology

Laboratory, Department of Entomology, Michigan State University.

iii

Dr. Jay E. McPherson (Southern Illinois University, Carbondale) and Deb Nelson
(USDA APHIS, Niles, Michigan) provided their expertise in identifying pentatomids and
gypsy moth parasitoids. Frank Sapio (Michigan Department of Natural Resources) and
Dr. Joseph Elkinton (University of Massachusetts, Amherst) provided soil used in our
introductions of E. maimaiga, and Fred Warner (MSU) tested soil samples for plant
parasitic nematodes. Dr. Ann Hajek (Cornell University, Ithaca, New York) was
extremely helpful with many aspects of my E. maimaiga work. Terry Davis (MSU)
assisted me with equipment, supplies and advice. The USDA Forest Service, State and
Private Forestry in St. Paul, Minnesota provided defoliation maps. Funding for this
research was provided by the McIntire-Stennis Cooperative Forestry Research Grant
program, the Michigan Department of Natural Resources, and the Professor Hutson
Student Research Award from the Department of Entomology at MSU.

I also thank my fellow graduate students (Tim Work, Bryan Bishop, Cathy Herms,
Beth Dankert, Amy Christensen, Nate Siegert, John Wise, Grzegorz Krawczyk and Bob
Kriegel) for helping to make this an enjoyable experience. Last, but certainly not least, I
wish to express a special thanks to Eileen Eliason for her friendship and constant

encouragement throughout my time at MSU.

iv

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................ vii
LIST OF FIGURES ........................................................................................................ ix
INTRODUCTION ........................................................................................................... 1
CHAPTER 1
Predicting Gypsy Moth (Lepidoptera: Lyrnantriidae) Defoliation: A Comparison
of Three Egg Mass Survey Methods ............................................................................. 4
Introduction .............................................................................................................. 4
Methods .................................................................................................................... 7
Study site ............................................................................................................... 7
Egg mass surveys ................................................................................................... 8
Defoliation ............................................................................................................. 9
Pathogens ............................................................................................................ 10
Statistical analyses ................................................................................................ 10
Results .................................................................................................................... 12
Comparison of egg mass measurements ................................................................ 12
Egg mass counts and defoliation ........................................................................... 12
Time .................................................................................................................... 15
Entomophaga maimaiga and NPV ....................................................................... 15
Discussion ............................................................................................................... 17
CHAPTER 2
Estimation of Seasonal Leaf Area Index with a Sunfleck Ceptometer .......................... 35
Introduction ............................................................................................................ 35
Methods .................................................................................................................. 37
Study site ............................................................................................................. 37
Data analysis ........................................................................................................ 38
Results .................................................................................................................... 39
Discussion ............................................................................................................... 4]
CHAPTER 3
Evaluation of Gypsy Moth Biological Control at the W. K. Kellogg Experimental
Forest, Michigan, Including Introductions of Two Natural Enemies ............................ 52

Introduction ............................................................................................................ 52

Methods .................................................................................................................. 53
Biological control introductions ........................................................................... 54
Entomophaga maimaiga bioassay - 1995 ............................................................. 56
Entomophaga maimaiga bioassay - 1996 ............................................................. 56
Larval sampling .................................................................................................... 58
Mortality diagnosis ............................................................................................... 58
Statistical analyses ................................................................................................ 59

Results .................................................................................................................... 60
Parasitoids, predators and pathogens .................................................................... 61
C otesra melanoscela ............................................................................................ 63
Entomophaga maimaiga ...................................................................................... 64

Discussion ............................................................................................................... 66

SUMMARY .................................................................................................................. 77

APPENDIX 1 - Record ofDeposition ofVoucher Specimens 79

LIST OF REFERENCES ............................................................................................... 81

vi

Chapter 1.

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

LIST OF TABLES

Coefficients of Pearson's product-moment correlation (r) and Spearman's
rank correlation (r.) among gypsy moth egg mass (EM) density, EM length,
and the percentage of new EMs at 32 sites in the Huron-Manistee National
Forest in Michigan. Significance levels: * = p<0.05; ** =p<0.01; NS = not
significant.

Coefiicients of Spearman's rank correlation for comparisons between gypsy
moth egg mass (EM) variables and defoliation at 32 sites in the Huron-
Manistee National Forest in Michigan. Egg mass density was measured in
each plot using three survey methods. Significance levels: * = p<0.05; ** =
p<0.01; NS = not significant.

Potential control decisions for 32 study stands based on gypsy moth egg
mass (EM) density, and the actual defoliation that occurred. These
thresholds are commonly used by gypsy moth managers to determine if
control is needed to prevent defoliation from exceeding 30%. The protocol
for the lOO-tree plot method is described in the text of this paper.

Average time needed to conduct gypsy moth egg mass (EM) surveys using
fixed-radius and lOO-tree plots at various population levels. Egg mass
density categories for the two methods are not directly comparable because
the methods sample different areas.

Spearman's nonparametric rank correlation coefficients among the percent of
gypsy moth larvae killed by the pathogens E. maimaiga and NPV, egg mass
(EM) density counts from three survey methods, and subsequent defoliation.
Egg mass surveys were conducted in the spring of 1995, and pathogens and
defoliation were evaluated in the summer of 1995. n=28 stands.
Significance levels: * = p<0.05; ** = p<0.01; NS = not significant.

vii

Chapter 2

Table 1.

Table 2.

Table 3.

Chapter 3

Table 1.

Table 2.

Table 3.

Summary of stand characteristics for the nine sample stands at the W. K.
Kellogg Forest in Kalamazoo Co., MI. Basal area, average DBH, density,
and height were measured in 1993 for all trees 2 2.5 cm DBH.

Summary of average leaf area index (LAI) and coeflicient of variation (CV)
at fiill leaf expansion for the nine sample stands at the W. K. Kellogg Forest
in Kalamazoo Co., MI. Average LAI and CV were computed using stand
LAIs for sampling periods between 11 June and 4 Sept. each year. LAI
includes foliage only.

Summary of woody area index (WAI) and maximum leaf area index (LAI)
from the nine sample stands at the W. K. Kellogg Forest in Kalamazoo Co.,
MI. Maximum LAI includes leaves only.

Summary of stand characteristics for the nine sample, stands at the W. K.
Kellogg Forest in Kalamazoo Co., lvfl. Basal area, average DBH, density,
and height were measured in 1993 for all trees 2 2.5 cm DBH.

Average pheromone trap and egg mass survey counts (moths/trap and egg
masses/ha, respectively). "Forest-wide" counts include 68 traps and 33 egg
mass survey locations. "Treatment plots" only include the nine biocontrol
plots. Egg mass counts are for the year in which the eggs were laid.

Total rainfall and deviation from the 50-year average (in cm) for May and
June 1994 and 1995 at the W. K. Kellogg Experimental Forest, Michigan.

viii

Chapter 1

Figure 1.

Figure 2.

LIST OF FIGURES

Linear regression of gypsy moth egg mass (EM) densities based on fixed-
radius and lOO-tree plot surveys of 32 sample plots in the Huron-Manistee
National Forest, MI. The slope and intercept were significantly different
from 0 (F=42.1,p<0.0001, SE=0.661; t=-2.42,p=0.022, SE=O. 172,
respectively).

Linear regression of gypsy moth egg mass (EM) densities based on fixed-
radius plot and timed walk surveys of 27 sample plots in the Huron-

. Manistee National Forest, MI. The slope and intercept are significantly
. different from 0 (F = 19.3,p = 0.0002, SE=1285.2; t = -3.10,p < 0.005,

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

SE=5074.0, respectively).

Linear regression of gypsy moth egg mass (EM) densities based on 100-tree
plot and timed walk surveys of 27 sample plots in the Huron-Manistee
National Forest, M]. The slope is significantly different from 0 (F = 12.0,

p < 0.002, SE=3110.6), but the intercept is not (SE=1078.8).

Percentage of new gypsy moth egg masses versus egg mass density from the
fixed-radius plots. Defoliation was grouped into three categories. All
stands with over 30% defoliation had at least 84% new EM and 6563
EM/ha, as indicated by the dotted lines. n=32 sites.

Gypsy moth egg mass (EM) length versus EM density in the fixed-radius
plots. Defoliation was grouped into three categories. All stands with over
30% defoliation had mean EM lengths between 19.3 and 30.4 mm, and at
least 6563 EM/ha, as indicated by the dotted lines. n=31 sites.

Percentage of new egg masses versus egg mass length. Defoliation was
grouped into three categories. All stands with over 30% defoliation had at
least 84% new EM and mean EM lengths between 19.3 and 30.4 mm, as
indicated by the dotted lines. n=3l sites.

Time required to conduct gypsy moth egg mass surveys using fixed-radius
plots and lOO-tree plots by egg mass density. n=32 sites.

ix

Figure 8.

Chapter 2

Figure 1.

Figure 2.

Figure 3.

Chapter 3

Figure 1.

Figure 2.

Figure 3.

Average percentage mortality caused by Entomophaga maimaiga and NPV
in gypsy moth cadavers collected at 28 study sites in mid-July, 1995 in the
Huron-Manistee National Forest, MI. Error bars indicate SEM. Average
percentage of mortality varied between the 2 defoliation levels for E.
maimaiga (p=0.004) and NPV (p=0.001). Percentage mortality varied
between the 2 pathogens for both low (p<0.001) and high (p=0.033)
defoliation. Tests were conducted with the nonparametric Wilcoxon two-
sample test.

Plot layout for ceptometer study. Each ‘x’ marks a sampling point. Stand
attributes (basal area, diameter, height, and density) were measured in four
0.01 ha plots indicated by the circles.

Leaf area index (foliage only) during 1994 and 1995 by individual stand.
Stand identification numbers are in the upper left comer of each graph.

Frequency distribution of time required for field sampling of one plot (40
ceptometer readings) in 1994 and 1995 (n=141).

Total counts of live gypsy moth larvae under burlap bands in the C.
melanoscela release plots at Kellogg Forest in 1994 and 1995. Plot
identification numbers are in the upper left corners. n = 18 bands.

Total counts of live gypsy moth larvae under burlap bands in the E.
maimaiga release plots at Kellogg Forest in 1994 and 1995. Plot
identification numbers are in the upper left comers. n = 18 bands.

Total counts of live gypsy moth larvae under burlap bands in the control
plots at Kellogg Forest in 1994 and 1995. Plot identification numbers are in
the upper left corners. n = 18 bands.

INTRODUCTION

The gypsy moth (Lymantria disparL.) is one of the most serious defoliators of
hardwood trees in the eastern United States. Outbreaks of this exotic insect often result in
heavy defoliation over large geographic areas, sometimes in excess of five million ha
annually (Butalla 1996). Defoliation can annoy recreationists, and can damage or kill trees
through growth loss, decreased vigor, and increased susceptibility to other stresses
(Kulman 1971, Wargo 1977, Herrick and Gansner 1987). Management of the pest can be
difficult. Population densities may vary within each infested stand, and estimates of local
population density therefore may not accurately predict the amount of subsequent
defoliation. Dispersal of early instars and mortality caused by predators, parasitoids, and
pathogens can contribute to inaccurate predictions of defoliation based on population
survey methods. Other management difficulties include introduction and evaluation of
biological control agents and quantification of defoliation.

This project was initiated because gypsy moth populations were expanding into
southwestern Michigan. Michigan State University’s W. K. Kellogg Experimental Forest
in Kalamazoo County provided an opportune site to evaluate some of the gypsy moth
management tools that are currently being used. Gypsy moth had been detected in this

area for several years, but had not yet caused noticeable defoliation. Research was also

2

conducted in the Huron-Manistee National Forest in northwestern Lower Michigan where
gypsy moth populations had gone through the initial outbreak in the early 1990s

Population surveys measuring egg mass density and quality are frequently used to
predict defoliation. Numerous protocols have been developed, each with various
advantages and disadvantages (Ravlin et al. 1987). An ideal survey would be easy to
conduct, require little time, and provide accurate predictions of defoliation. In Chapter 1,
we present a comparison of 3 egg mass survey methods, discuss their time requirements,
and evaluate their accuracy in predicting defoliation. The first method we evaluated was
the fixed-radius plot (Kolodny-Hirsch 1986), which is generally considered to provide the
most accurate and precise estimate of actual egg mass density (Kolodny-Hirsch 1986,
Liebhold et al. 1991). We also tested the timed walk (Eggen and Abrahamson 1983), a
method used by county gypsy moth suppression personnel in Michigan until 1997. The
third method was the lOO-tree transect, a method developed in Michigan that may
combine the accuracy and precision of the fixed-radius plots with the time efficiency of the
timed walks (Chilcote et al. 1994). These methods were evaluated in stands with low to
high gypsy moth populations in the Huron-Manistee National Forest area.

Gypsy moth defoliation can be quantified in many ways. Aerial surveys are useful for
determining broad categories of defoliation over large areas. Stand-level defoliation may
be estimated visually, but this provides only qualitative estimates that are subject to
observer bias. In Chapter 2, we describe the use of a Sunfleck Ceptometer to quantify leaf
area between bud break and leaf fall in 9 hardwood stands. The objective was to evaluate

consistency of the measurements and determine the potential of the ceptometer for

3

measuring changes in leaf area caused by gypsy moth defoliation. This work was
conducted at the Michigan State University W; K. Kellogg Experimental Forest.

Gypsy moth management often includes the release of biological control agents to
regulate gypsy moth populations without the environmental effects associated with
conventional insecticides. In Chapter 3, we present results from a study designed to
establish the braconid wasp C otesia melanoscela (Ratzeburg) and the fungus
Entomophaga maimaiga Humber, Shimazu and Soper at Kellogg Forest, an area only
recently invaded by gypsy moth. We also identified other gypsy moth natural enemies that
were already present in the area. These natural enemies include previously established
biocontrol agents that had moved along with the gypsy moth, and native generalist
parasitoids and predators that were already present.

Much of the gypsy moth research conducted to date and most management strategies
were developed in the northeastern United States where gypsy moth has been present
since 1869. However, research findings and management options have not been
adequately evaluated in recently infested areas like Michigan where different climate and
cover types may cause changes in gyspy moth population dynamics. The goal of this
project was to provide information that will help improve gypsy moth management in

Michigan and other North Central states.

CHAPTER 1

Predicting Gypsy Moth (Lepidoptera: Lymantriidae) Defoliation: A Comparison

of Three Egg Mass Survey Methods

INTRODUCTION

The gypsy moth, Lymantria dispar L., is a major defoliator of hardwood trees in the
eastern United States. It is a pest not only in forests, but also in forested rural and urban
areas. Defoliation has serious impacts on trees, including growth loss, decreased vigor,
increased susceptibility to other stresses, and even mortality (Kulman 1971, Wargo 1977,
Herrick and Gansner 1987). Suppression of high gypsy moth populations with biological
or chemical insecticides is often necessary to keep damage and annoyance at acceptable
levels. Defoliation in the United States between 1980 and 1995 ranged from 293,000 to
5.2 million ha (Butalla 1996). Annual suppression costs during this time period were as
high as $22.5 million (Butalla 1996).

Decisions on the need for gypsy moth suppression using pesticides are based largely
on population surveys used to predict subsequent defoliation. Inaccurate population
estimates can lead to overspraying and associated economic and environmental impacts.
Forest managers prefer to survey overwintering egg masses because they are present

longer than other life stages and because sampling can be completed in time to plan

5
suppression programs (Fleischer et al. 1991). Egg mass density thresholds and

information on population trends (increasing or decreasing) are typically used to delineate
treatment areas (Onken et al. 1996).

At least 13 different protocols have been used by managers in the United States to
sample gypsy moth egg masses (Ravlin et a1. 1987). Two common methods are the timed
walk and the fixed-radius plot. In the timed walk (Eggen & Abrahamson 1983), the
observer walks in a predetermined direction for 5 min (a usual distance of 0.16 km),
counting all visible egg masses on the trunks and branches of trees. Then the observer
reverses direction and repeats the survey back to the starting point and averages the 2
counts. Tables derived from regression equations are used to convert counts into density
estimates of egg masses per acre (Eggen & Abrahamson 1983). This method is easy to
learn and use and has a relatively low cost (Ravlin et al. 1987). However, the association
between timed walk counts and actual egg mass density is confounded by differences
among observers and habitats (Fleischer et al. 1991, Liebhold et al. 1991).

Observers using the fixed-radius plot method count egg masses within a circular plot
usually 0.01 ha (1/40th acre) in size (Kolodny-I-Iirsch 1986). Egg masses are counted on
ground debris and standing trees within the plot. This method is more labor-intensive than
the timed walk, particularly in high populations, and often requires binoculars for counting
masses high in tree canopies (Liebhold et al. 1994). However, the fixed-radius plot is
considered the most accurate and precise method for estimating egg mass density
(Kolodny-Hirsch 1986, Liebhold et a1. 1991). Researchers commonly use fixed-radius
plots to determine egg mass density, but in operational projects there may not be adequate

resources to thoroughly survey large areas using this method.

6

Chilcote et a1. (1994) recently developed an alternative egg mass survey method in
Michigan based on a linear transect of 100 trees. The observer counts new egg masses on
the lower 2 m of tree holes that are at least 5 cm in diameter. New and old egg masses
can be easily distinguished since all egg masses are within reach of the observer. Chilcote
et al. (1994) developed a two-step procedure for decision-making based on a goal of
preventing more than 37% defoliation (the defoliation level of concern for many
managers). In step one, control is not warranted if the average number of new egg masses
per tree is less than or equal to 0.2, and control is recommended if the average is 2.0 or
greater. If the average number of egg masses per tree is between 0.2 and 2.0, a regression
model is used to predict defoliation and make the control decision. This model requires 3
variables: the proportion of the trees that are oak (Quercus spp.), average egg mass
length, and the ratio of the number of new to old egg masses. Although this method has
not yet been integrated into an operational program, it may reduce time needed to survey
large areas while retaining a high level of accuracy.

Associations between egg mass density and subsequent defoliation are often
complicated by the presence of pathogens. The nuclear polyhedrosis virus (NPV) has
historically been a major factor regulating gypsy moth in the United States, typically
causing crashes in outbreak-level populations. Another pathogen is the fungus
Entomophaga maimaiga Humber, Shimazu and Soper. This larval pathogen mysteriously
appeared in outbreak gypsy moth populations in 7 northeastern states in 1989 (Andreadis
and Weseloh 1990, Hajek et al. 1990a). In 1991, Smitley et al. (1995) introduced this

pathogen at several sites in northwestern Michigan. Favorable weather and subsequent

7

introductions have resulted in widespread distribution of E. maimaiga in the northeastern
region and in Michigan.

Our first objective was to evaluate the accuracy and time requirements of 3 egg mass
survey methods, the 100-tree plot, fixed-radius plot, and timed walk in oak forests of
Michigan’s lower peninsula. We also examined the associations between defoliation and
various egg mass variables. Our second objective was to evaluate gypsy moth mortality
caused by Entomophaga maimaiga and NPV, and how these pathogens differed among

gypsy moth populations.

METHODS
Study site

Study sites were established in 32 stands in the western half of the Huron-Manistee
National Forest. These stands are located in the northwestern part of Michigan's lower
peninsula in Manistee, Mason, Lake, Newaygo, and Oceana counties. Each study stand
met the following criteria: the terrain was flat to gently rolling, there was no history of
Bacillus thuringiensis application, at least 60% of the basal area was mature oak, and
each stand was at least 10 ha in size. This area had been through at least 1 gypsy moth
outbreak prior to this study (Defoliation maps, Michigan Department of Natural
Resources).

Gypsy moth defoliation maps from the Michigan Department of Natural Resources
and the USDA Forest Service, State and Private Forestry in St. Paul, Minnesota were
used to generate a list of 100 potential study stands. Stands across a range of gypsy moth

population levels were randomly selected from this list and sampled after verifying that

8

they met the selection criteria. Dominant overstory species in the study stands were white
oak (Quercus alba L.), red oak (Q. rubra L.) and black oak. (Q. velutina Lam.) The

understory plants ranged in height from 0.1 to 3.0 m, and varied in species and abundance.

Egg mass surveys

One study site was established in each stand at least 50 m from the edge of the stand
or road to avoid edge effects. Surveys were conducted between 20 December 1994 and
12 May 1995 before egg hatch occurred. The 3 egg mass surveys were conducted in
random order at each site. The 0.01 ha fixed-radius plot and 100-tree plot were centered
at a wooden stake marking each study site, and a timed walk was conducted in a random
direction from the stake.

Fixed-radius plots. We carefully examined all trees within the 0.01 ha plot for egg
masses, scanning the entire tree with binoculars. Logs and branches on the ground within
the plot were overturned in search of egg masses.

Timed walks. After randomly choosing a direction, we tallied egg masses with a hand
counter along a transect for 5 min. We repeated the count along the route back to the
starting point and then averaged the 2 counts (Abrahamson 1987).

IOU-tree plots. We modified the 100-tree transect method of Chilcote et al. (1994)
into a circular plot of the closest 100 trees around a central point so that counts could be
expressed on an area basis. We began from the tree at the center of the plot and walked in
a circular path, spiraling outward from the center until we had examined 100 trees that
were at least 4 cm in diameter. A hand counter was used to count the new egg masses on
the lower 2 m of each tree bole, and each tree was marked with chalk to ensure that it was

not examined twice. Old egg masses were tallied at the same time and used to compute

9

the percentage of new egg masses for the site. This percentage was then used to adjust
counts from the fixed-radius plots and timed walks, where all egg masses were counted.
To determine the ground area of each lOO-tree plot, we measured the distance between
the plot center and the farthest tree examined in each cardinal compass direction. The
mean of these 4 "pseudo-radii" was used to calculate the circular area of the plot. Egg
mass counts were then converted into a standard unit of egg masses per ha.

At each site we measured the length along the longest axis of the first 10-15 egg
masses encountered. The time needed to complete each survey was recorded as well.
Time needed to travel to the plot’center was the same for all sampling methods, so only

the time required to complete each survey was recorded.

Defoliation .

We assessed defoliation at each study site during peak defoliation in July 1995 after
most gypsy moths had pupated. Defoliation of trees within 30 m of the plot center was
visually estimated and assigned to 1 of 8 percentage classes: 0 to 20, 21 to 30, 31 to 40,
41 to 50, 51 to 60, 61 to 75, 76 to 90, and 91 to 100%. This follows the methods of
Carter et al. (1991). Estimates were made separately for overstory oaks, and for
understory oaks and witch-hazel (Hamamelis virginiana L.) shrubs. Oaks and witch—
hazel are preferred hosts for gypsy moth (Liebhold et al. 1995). The percent cover of
gypsy moth-preferred hosts in the understory was visually estimated as none/sparse,
common, or abundant. We also noted the presence or absence of live gypsy moth larvae
and pupae while doing defoliation estimates. If live larvae were found, we returned 2 wks
later to verify the original defoliation level estimates. If more defoliation had occurred, we

used the estimates from the later visit.

10

Pathogens

We collected the first 1 to 6 gypsy moth larval cadavers that we encountered on tree
boles at each site, depending on how common the cadavers were. Each cadaver was taken
from a different tree whenever possible. We collected a total of 108 cadavers at 28 sites
(mean = 3.9 cadavers per site); no cadavers were present in 4 sites. Cadavers were placed
into individual plastic containers, returned to the laboratory, and frozen for later analysis.

A sample of internal tissue and integument from each cadaver was smeared on a
microscope slide and stained with lactophenol cotton blue fungal stain (Poinar and
Thomas 1978). Slides were examined at 100X for E. maimaiga resting spores and conidia
and at 400x for NPV polyhedral inclusion bodies (Hajek and Roberts 1992). Head
capsule widths of approximately half of the cadavers were measured to determine the

frequency of each larval instar (Jobin et al. 1992).

Statistical analyses

Comparison of egg mass measurements. Linear regression (Sokal and Rohlf 1995)
was used to examine associations among egg mass length, percentage new egg masses,
egg mass densities from the 3 methods, and time needed to complete each survey. Since
egg masses were counted on all substrates (ground debris and trees) within the fixed-
radius plots, we considered counts from this method to be measures of absolute egg mass
density (Fleischer et al. 1991). Limited portions of the total substrate are searched with
the timed walks and 100-tree plots; therefore they provide measures of relative density.
Estimates of relative density (timed walks and 100-tree plots) were regressed as a linear
function of absolute density (fixed-radius plot) estimates to evaluate associations among

the 3 methods. The tables provided by Eggen and Abrahamson (1983) for converting

ll

timed walk counts into per ha densities are not considered valid when the average count
for the 2 walks is less than 20 egg masses and the area is not experiencing its initial
outbreak (Abrahamson 198 7). Five sites did not meet these criteria and were therefore
dropped from analyses that used timed walk counts. We examined associations among
egg mass variables using parametric (Pearson product-moment) and nonparametric
(Spearman’s rank) correlations, as apprOpriate (Sokal and Rohlf 1995). Data were
transformed [log10(x+1)] when necessary to meet the assumptions of normality and
homogeneous variances.

Egg mass measurements and defoliation. Spearman’s rank correlation was used to
examine associations among the 8 defoliation classes, egg mass measurements, and
abundance of understory vegetation. Defoliation categories were combined for further
analysis due to low sample sizes in some categories. Final categories used in analyses
were 0 to 30% and 31 to 100% defoliation. Forest managers often use the 30% threshold
as the defoliation level where control treatment becomes needed (Liebhold et al. 1994,
Onken et al. 1996). A t-test was used for each survey method to determine if mean egg
mass densities differed between sites that experienced low defoliation (0-30%) and sites
that experienced moderate to severe defoliation (31-100%).

Pathogens. We calculated the percentage of larval cadavers collected that were killed
by each of the 2 major pathogens in each site. We then examined the associations between
this variable and both defoliation level and egg mass density using Spearman’s rank
correlation. We used the nonparametric Wilcoxon two-sample test (Sokal and Rohlf
1995) to test for significant differences in pathogen infection rates within and between

sites with low defoliation and sites with moderate to severe defoliation. Contingency table

12

analysis (Sokal and Rohlf 1995) was used to determine if pathogen infection rates varied
among instars and between sites where pupae and live larvae were present or absent.
Significance of these models were tested with the likelihood ratio Chi-square ()8) test
(Sokal and Rohlf 1995). All significance tests were evaluated with alpha = 0.05. Analyses

were conducted using JMP statistical software (SAS Institute 1995).

RESULTS
Comparison of egg mass measurements

The lOO-tree plot egg mass densities explained more of the variation in fixed-radius
plot density estimates (R2=0.58) (Figure 1) than—did the timed walks (R2=0.44) (Figure 2).
Counts from the timed walks and 100-tree plots were also significantly associated with
each other (R2=0.32, Figure 3). Egg mass length varied inversely with egg mass density
estimated with each survey method (Table 1). The percentage of new egg masses was not

significantly associated with egg mass density or with length (Table 1).

Egg mass counts and defoliation

Defoliation maps from the Michigan Department of Natural Resources and the US.
Forest Service indicated that defoliation was less than 75% in our 32 study sites in 1993,
and was between 50 and 100% in 22 of our sites in 1994. During 1995, the year of this
study, 26 sites had less than 50% defoliation. All of the sites with over 50% defoliation in
1995 also had over 50% defoliation in 1994, according to defoliation maps. Defoliation of
susceptible host species in the understory (oaks and witch-hazel) was below 30% in 24
sites, between 31 and 75% in 6 sites, and over 7 5% in the remaining 2 sites in 1995.

Initial statistical analyses were conducted using the original 8 categories of defoliation.

l3

Overstory defoliation was significantly correlated with egg mass counts from the fixed-
radius and 100-tree plots but not with timed walk counts (Table 2). Understory
defoliation followed the same pattern; it was significantly correlated with the fixed-radius
and 100-tree plot counts but not with the timed walk counts (Table 2). The amount of
cover of susceptible host species in the understory (none/sparse, common, or abundant)
was not significantly associated with the amount of defoliation in either the understory or
overstory. Defoliation levels in the understory and overstory were positively correlated
with each other (r,=0.627, p<0.001).

We conducted further analyses to determine how well each method could be used to
distinguish between low (0 to 30%) and moderate to heavy (31 to 100%) defoliation. A t-
test was used to evaluate differences between mean egg mass densities in the 2 groups.
Differences between means were significant for estimates from the fixed-radius (t=3.60;
df=30; p<0.002) and 100-tree (t=3.54; df=30; p<0.002) plots, but not for the timed walks
(t=1.45; df=25; p=0.16). Although there was overlap in egg mass densities between the 2
defoliation levels for all 3 survey methods, the overlap was greatest in the timed walk
counts.

Percentage of new egg masses in a plot was correlated with overstory defoliation but
not with understory defoliation (Table 2). Egg mass lengths were not significantly
correlated with overstory or understory defoliation. Mean egg mass length i SEM was
25.2 mm i 1.2 and 23.7 i 1.8 for lightly defoliated stands (0 to 30%) and moderate to
severely defoliated stands (31 to 100%), respectively. Mean egg mass length was not

significantly different between the 2 defoliation levels (t=-0.65, df=29, p=0.52).

14

Figures 4 - 6 show various combinations of 3 potential defoliation predictor variables,
percentage new egg masses, egg mass length, and egg mass density from fixed-radius
plots. The high degree of overlap of data in these graphs makes it difficult to select a set
of variables that best predicts defoliation. Percentage new egg masses and egg mass
density appear to be the best combination of predictors (Figure 4). All stands with
defoliation over 30% had at least 84% new egg masses and 6563 egg masses/ha. Use of
egg mass length added little benefit over using 1 of the other variables alone (Figures 5
and 6).

We used common egg mass density thresholds along with counts from the 3 egg mass
survey methods to evaluate the 32 stands for gypsy moth suppression (Table 3),
simulating the decision-making process of a manager. The timed walks and fixed-radius
plots gave similar results, correctly recommending treatment for all the stands where
defoliation actually exceeded 30%. However, these methods recommended treatment for
17 to 20 stands where defoliation was less than 30% and treatment would be unnecessary.
We used the protocol in Chilcote et al. (1994) and Chilcote et al. (unpublished data) to
make control decisions based on our 100-tree plot data. Since tree species were not
recorded, we estimated that 60% of the trees in each stand were oak. Using the model,
control decisions were made with the first step in 14 of the 32 sites, with 1 incorrect
decision. The other 18 decisions were made in the second step using the regression
model, with 10 incorrect decisions. All incorrect decisions erred on the side of

recommending treatment when it would not have been necessary.

15

Time

Timed walk surveys took the least amount of time, always requiring 10 min. Fixed-
radius plots took an average of 18.1 min to complete while 100-tree plots required the
most time, averaging 30.0 min each. On average, 3.5 min (12%) of the total time required
for the 100-tree plots were used to measure the pseudo-radii. Counting old egg masses
during the 100-tree surveys did not increase the time required to any significant degree.
As expected, regressions showed that egg mass density influenced the amount of time
needed for the fixed-radius plots (R2=0.38, F=18.4, df=31, p<0.001) and 100-tree plots
(R2=0.40, F=19.9, df=3 1, p<0.001) (Figure 7). The difference in time requirements .

between the 2 methods increased with increasing egg mass density (Table 4).

Entomophaga maimaiga and NPV

We found gypsy moth cadavers in 28 of the 32 stands we surveyed. E. maimaiga was
the primary mortality agent for 65 of the 109 cadavers collected. NPV killed 35 larvae,
and 9 larvae died of unknown causes. The unknown deaths may have been caused by
parasitoids, predators, stress, pathogens at undetectable levels, or unknown factors. E.
maimaiga was found in 21 of the 28 sites where cadavers were collected and NPV was
found in 16 of the 28 sites. We determined the instar of 69 of the 109 larvae. Most were
sixth instars (74%), 22% were fifth instar, and 4% were fourth instars. Contingency table
analysis indicated that the primary cause of mortality did not vary depending on instar
(x2=0.333. p=0.85).

Percentage mortality caused by each pathogen was significantly associated with
defoliation and preseason egg mass density (Table 5, Figure 8). As expected, NPV was

primarily associated with high gypsy moth populations. We found NPV in all sites that

16

had over 30% defoliation, but in only 8 of 20 sites with low defoliation. E. maimaiga had
a negative association with defoliation. It was found in 17 of the 20 stands‘with low
defoliation where we found cadavers, and in half of the high defoliation stands. Amount
of mortality caused by either pathogen was not significantly correlated with egg mass
length or the percentage of egg masses that were new (p>0.27).

We observed live gypsy moth larvae in 22 of the 28 stands where we collected
cadavers. Stands where no live larvae were observed had significantly higher infection
rates of E. maimaiga than NPV (x2=7.022, p=0.008). The 6 stands where we did not find
live larvae all had E. maimaiga infections rates of 75 to 100%. It follows that sites where
live larvae were observed tended to have higher levels of NPV than stands with no live
larvae (x2=3.969, p=0.046). Contingency table analysis also indicated that defoliation
category and the presence/absence of live larvae were independent (x2=0.575, p=0.45).

We observed gypsy moth pupae in 11 of the 28 stands where we collected cadavers.
As expected, pupae were more commonly found in stands with high defoliation
(12:11.476, p<0.001). High infection rates of E. maimaiga tended to occur more
frequently in stands where no pupae were found (x2=4‘.550, p=0.033). NPV infection
rates did not vary between stands where we did or did not observe pupae (xz=l .226,
p=0.27). Also, there was no significant association between the presence/absence of
pupae and live larvae (x2=1.797, p=0. 18).

Percentage mortality attributed to E. maimaiga and NPV are measurements of the
relative abundance of these 2 pathogens in each stand, but do not reflect the percentage of
all larvae in the stands that were killed by these pathogens. These results should be

interpreted cautiously since the abundance of pupae and live larvae were not quantified in

17

each stand, although our observations are based on large amounts of time spent in the

stands.

DISCUSSION

Egg mass densities determined by the 3 egg mass survey methods are not directly
comparable because different areas are searched with each method. An observer using the
fixed-radius plot method searches for egg masses in the canopies and under logs and
rocks. The lOO-tree plots include counts from only the first 2 m of tree boles. Egg
masses counted during timed walks are primarily on the boles and undersides of major
branches. However, the tables for converting timed walk counts to egg masses per acre
were developed by regressing timed walk counts on counts from fixed- and variable-radius
plots (Wilson and F ontaine 1978), which include counts taken on trees and the ground
(Eggen and Abrahamson 1983). Egg mass density estimates from the timed walks
underestimated those from the fixed-radius plots by an average of 16% in this study.
Therefore, thresholds used for predicting defoliation will vary depending on the survey
method used.

Liebhold et al. (1991, 1994) concluded that the timed walk should not be used for
estimating gypsy moth egg mass densities because of low precision and high susceptibility
to observer bias. Kolodny-Hirsch (1987) and Liebhold et al. (1991) considered the fixed-
radius plot method to be the most cost-efficient and precise sampling method for
determining egg mass densities in forest stands. In our study, egg mass density estimates
from the timed walks and 100-tree plots were both significantly associated with fixed-

radius plot counts (R2=0.44 and 0.58, respectively).

18

The next question we addressed concerned relative accuracy of the 3 methods in
predicting defoliation. Gypsy moth managers must be able to accurately delineate areas
where defoliation will be high so that appropriate actions can be taken. Suppression
programs are very expensive; over $64 million was spent between 1991 and 1995 under
the USDA Forest Service Cooperative Suppression Program (Butalla 1996). Although
egg mass surveys are the preferred method for predicting defoliation, results can be
extremely variable. Gypsy moth managers tend to be risk-averse when determining
treatment thresholds to minimize potential conflicts with homeowners and forest
recreationists.

Correlations and t-tests showed that fixed-radius plot counts were significantly
associated with defoliation, while timed walk counts were not. But when counts from
these 2 methods were used along with egg mass density thresholds to evaluate stands for
suppression, results were very similar. If our results from fixed-radius plots or timed
walks had been used to evaluate the study stands for treatment, much unnecessary
spraying would have occurred. The 100-tree plot method would have provided better
results in planning a suppression program for our stands than would the timed walks or
fixed-radius plots, using commonly accepted thresholds. Treatment thresholds were
inappropriate, partly due to the declining gypsy moth populations. Maximum defoliation
typically occurs just before peak egg mass density (Williams et al. 1991). Therefore, egg
mass counts may indicate that the population is still growing, even though defoliation may
have just peaked. If surveys are conducted at peak egg mass density, then treatments

applied the following spring will probably exceed what is actually needed will be after

19

peak defoliation has already occurred. Also, all early research has been conducted before
E. maimaiga became a significant factor in gypsy moth population dynamics.

Many gypsy moth managers like the timed walk survey because it is quick, easy for
observers to learn, and each walk covers a lot of area. However, counts can only be
converted into estimates of egg masses per ha using the tables developed in New York
(Eggen and Abrahamson 1983). We also observed several problems with the timed walk
in the field. Egg masses were more apparent on certain tree species. For example, they
were very obvious on smooth-barked species like bigtooth aspen (Populus grandidentata
Michx.) and dark-barked tree species like black oak, but difficult to see on trees like white
oak, which have flaky bark similar in color to the egg masses. The timed walk method
was also more strongly influenced by the position of the sun. It was difficult to see egg
masses when our transect faced the sun during a timed walk, whereas the other methods
enabled us to search the trees from a slightly different angle. The timed walk was also
more hazardous for the observer, who had to continually search tree crowns while walking
over stumps and logs and through brush.

While the fixed-radius plot method also is easy to learn, it does require more time than
a timed walk. In areas of low egg mass densities it required an average of only 5 min
more time than the timed walks. The difference increased in high populations, where the
fixed-radius plot took over 13 min longer than the timed walk (Table 4).

Another problem with the fixed-radius plot method was that the egg mass density
estimate was biased upward when a plot had an individual tree with an extraordinary

number of egg masses. This is, of course, why mean egg mass density for a stand should

20

be determined by sampling multiple plots. The 100-tree plot was less affected by this
problem because using 100 boles diluted the effects of individual trees.

Egg mass density estimates from the 100-tree plots and the fixed-radius plots were
comparable in their association with defoliation, but the 100-tree plot method took nearly
12 min longer to complete. The major drawback to the lOO-tree plot method was the time
factor, requiring an average of 30 min per survey. This is not acceptable for forest
managers, who have limited time and resources for conducting surveys over large areas.
Chilcote et al. (1994) and Chilcote et al. (unpub. data) found that accuracy in predicting
defoliation was as good using 25 trees (89.8%) as it was using 100 trees (86.2%). Using
25 trees may make this method more feasible for use in gypsy moth management
programs. The 100-tree method also eliminates the need to determine the percentage of
the egg masses that are new, since the observer can closely examine all masses and ignore
the old ones. The fixed-radius plot and timed walk methods require this extra survey
when both types of egg masses are present. This time savings can be substantial;
surveying 100 egg masses to determine the percent of new and old took 4 to 16 min,
depending on the population density.

The average area encompassed by the 100-tree plots was 0.09 i 0.006 ha (0.23 i
0.014 acres, mean .+_ SE), 9 times larger than the fixed-radius plots. Raw counts of new
egg masses per 100 trees were highly associated with the counts expressed in units of egg
masses per ha (R2=0.93, F=430.1, df=3 1, p<0.0001). This suggests that measuring the
area of each lOO-tree plot may be unnecessary, which would save an average of 3.5 min
per plot. Counts of new egg masses per 100 trees were as highly correlated with

defoliation and other egg mass variables as were the counts after conversion to egg masses

21
per ha. Using the transect version of the method (Chilcote et al. 1994) may better

represent the stand because it covers a longer linear area, but may be more prone to
observer bias if the observer tends to orient toward big trees or trees with many egg
masses. Since only tree boles are searched in the lOO-tree plot method, it may also
underestimate egg mass density in low populations, since a higher proportion of egg
masses in low populations tend to be laid in concealed locations on the ground (Eggen and
Abrahamson 1983, Skaller 1985). Whether this error is enough to result in unexpected
defoliation is unknown.

Predicting defoliation from egg mass counts is confounded by a variety of factors that
affect gypsy moth eggs and larvae, including pathogens, parasitoids, weather, and the
presence and distribution of host trees and shrubs. Fourteen study stands had estimates of
over 2471 egg masses per ha ( 1000 masses/acre) but received less than 30% defoliation.
Cadavers killed by E. maimaiga and NPV were found in nearly all of these stands, with
the majority of the cadavers infected by the fungus. These pathogens likely had a big
impact on larval populations and subsequent defoliation.

Diagnosing E. maimaiga and NPV in cadavers from the study sites provided an
indication of relative infection rates among sites. As expected, pathogen infection rates
were associated with population levels as measured by egg mass density and defoliation.
Larval mortality due to NPV was positively correlated with defoliation and egg mass
density. This is typical of the density-dependent nature of the virus, which typically is
more prevalent when larvae are at high density (Wallner and McManus 1981). Mortality
rates caused by E. maimaiga were negatively correlated with defoliation and egg mass

density. Since NPV is not usually prevalent in low density populations, it was not

22

suprising that E. maimaiga killed a higher percentage of the larvae found in these stands.
Also, the fiingus is not as dependent on high gypsy moth larval density as the virus, and
high infection rates have been noted in low density gypsy moth populations (Hajek et al.
1996b). All of the stands where no live larvae were found had high (at least 75%) cadaver
infection rates of the fungus relative to NPV, and most of these stands also had low
defoliation. Stands where pupae and live larvae were not observed usually had greater
than 50% infection of cadavers by E. maimaiga. E. maimaiga may have caused so much
mortality in low-density populations that live larvae could not be easily found.

We initially considered that the association of E. maimaiga with defoliation and egg
mass density was related to the spatial distribution and spread of the fungus since its
original introduction. But E. maimaiga was found throughout the 5400 square km study
area. All sites where we did not find E. maimaiga were within 5.2 km of a site where it
was found. Although NPV was likely present in all study stands, we detected it in fewer
stands than E. maimaiga.

Since this study was conducted while most gypsy moth populations in the area were
declining, one must be careful in extrapolating results to healthy, growing populations.
More research is needed to determine how egg mass variables can be used most effectively
to predict defoliation across varying population levels. Also, a better understanding of
how pathogens subsequently affect population density and defoliation is needed. This will

become more important as E. maimaiga spreads across the range of gypsy moth.

23

Table 1. Coefficients of Pearson's product-moment correlation (r) and Spearman's rank
correlation (r,) among gypsy moth egg mass (EM) density, EM length, and the percentage
of new EMS at 32 sites in the Huron-Manistee National Forest in Michigan. Significance
levels: * =p<0.05; ** =p<0.01; NS = not significant.

 

Correlation of EM density with:

 

 

EM length Percentage new EMS
Survey Methods (r) (r,)
Fixed-radius plots -0.544* NS
100-tree plots -0.391* NS
Timed walks -0.394* NS

 

Table 2. Coefficients of Spearman's rank correlation for comparisons between gypsy moth
egg mass (EM) variables and defoliation at 32 sites in the Huron-Manistee National Forest
in Michigan. Egg mass density was measured in each plot using three survey methods.
Significance levels: * = p<0.05; ** = p<0.01; NS = not significant.

 

Correlation of EM variables with:

 

 

Egg mass variables Overstory defoliation Understory defoliation

EM density 0.564” 0.472**
(Fixed-radius plots)

EM density 0.701** 0.640**
(100-tree plots)

EM density NS NS
(Timed walks)

EM length NS NS

Percentage new EMS 0366* NS

 

24

5188855 8285 83 26: 2:63 85 8.8% 38858 2828 $5. A
{com mcmvoooxo 88:83 was

26; 8822: 2:63 was wowcoEEooE :82 96: 8c 2:63 8:5on 75:8 823, macaw 38858 5:28 mu; N
#8: .E 8 88.30 com

32a 85-2: .8”— .Gao: .3 8 E615 8m 32a 368-886 Ea 3:55 BE: wEm: mac—058:: E258: Sm _

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

w E o 3 32.. 8.5-8—
” M: o o 88: EN
32.— 353-85,.—
» om o a. son 22
w 2 o a Soc 32
9:; 85:.
w 2 _ o m soc E
985,... ao .oz 92% co .oz 92% Lo .2 $853.8 .oz 6825 32m 850:. 3&5
see; m ”some N387 _ m some 328::
zeta—86c 333w 853—826 RENE €258;
cacao—niece: .2280 86:98:8qu .8280 :2
898 $5 Lo ~on 65 E

contact E 852: 83 85-2: 2: he :86on 2:1 .°\oom ch888 Set 88:88 E965 8 886: 3 35:8
3 655.8% 8 muowmcmfi 59: $93 3 wow: 38888 2m 8356:: 08: 1.. .8558 :2: 88:88 328
2: 98 56:26 32.3 mme wwo 59: >83 so 8mg 38% >25 mm com $538.6 35:8 3388; .m 28,—.

25

Table 4. Average time needed to conduct gypsy moth egg mass (EM) surveys
using fixed-radius and 100-tree plots at various population levels. Egg mass
density categories for the two methods are not directly comparable because the
methods sample different areas.

 

 

 

 

 

 

Survey EM density range Time range (minutes) Mean time i SE
method EM/ha (EM/ac) Min. Max. minutes
Fixed-radius 0 - 1,236 8.0 25.5 15.1 :I: 2.7
(0 - 500)
1,237-2,47l 11.3 20.0 15.5i 1.5
(501 - 1,000)
2,472-24,710 14.0 23.5 17.63: 1.0
(1,001 - 10,000)
>24,710 15.8 31.8 235$ 1.9
(>1CL000)
100-tree 0- 124 17.0 31.5 22.9i2.5
(O - 50)
125 - 494 19.3 32.8 24.9 i 1.4
(51 - 200)
495-1,236 17.5 50.0 31.5:l:3.7
(201 - 500)
>1,236 26.3 52.5 37.0 i 2.7

(>500)

 

26

 

tamed immod mZ *aoomd acmvd >mZ

 

 

 

 

 

 

amovd- aawvcd- m2 12:6- tamed- SMEEEE m
b08833 beefing: 8:33 BEE. 32: 25-2: $2: 368.85.: :3 838 9:358 °\o
53.3.89: ”.3 858.58.. bane: Zn..— 888.»...—

 

aaéewe so: u mz :31 u ......

”3.on n a ”296— oogucmcwmm .853 mm”: .33 :6 868% 65 E 88236 26>» 88:86:

can meowofima :8 .32 :o wctam 65 E 888:8 whoa mxotsm mmmE wwm .:o:m:£oe Eon—833m :8
£852: taut—5 00E: :8: 3:58 588: :25 3.2: wwo .>mZ can emanates .m meowofima 2: 3 8:3
outa— fioE >83 :6 “882: 2: 98:8 $86568 8:32.80 :52 0508888: 988.8on .m 633.

27

 

 

 

 

 

4.0 - 0
y=-l.600+1.117x -
3.5 _ R2 = 058
O

5
Li]
5'3
.2
3
O
'5.
8
l:

0.5 -

0'0 4L. I I I I

2.5 3.0 3.5 4.0 4.5 5.0

Fixed-radius plots (log10 EM/ha)

Figure 1. Linear regression of gypsy moth egg mass (EM) densities based on fixed-
radius and lOO-tree plot surveys of 32 sample plots in the Huron-Manistee National
Forest, MI. The slope and intercept were significantly different from 0 (F=42. 1,
p<0.0001, SE=O.66]; t=-2.42, p=0.022, SE=O. 172, respectively).

28

 

 

 

 

15000 - y=-15725.3+5652.7x ' .
R2=0.44
12000 -
a?
E
E9, 9000 -
m
at
'a
3
B 6000 -
.§
[-
3000 -
0
2.5

 

Fixed-radius plots (loglo EM/ha)

Figure 2. Linear regression of gypsy moth egg mass (EM) densities based on fixed-
radius plot and timed walk surveys of 27 sample plots in the Huron-Manistee National
Forest, MI. The slope and intercept are significantly different from 0 (F = 19.3, p =
0.0002, SE=1285.2; t = -3.10, p < 0.005, SE=5074.0, respectively).

29

 

15000 7 y=-4104.5 +3741.3 x '
R2=0.32 W

1

12000

9000-

6000‘

Timed walks (EM/ha)

 

 

 

0 I I I l
1.0 1.5 2.0 2.5 3.0 3.5 4.0

 

loo-tree plots (log10 EM/ha)

Figure 3. Linear regression of gypsy moth egg mass (EM) densities based on lOO-tree
plot and timed walk surveys of 27 sample plots in the Huron-Manistee National Forest,
M1. The slope is significantly different from O (F = 12.0, p < 0.002, SE=3110.6), but
the intercept is not (SE=1078.8).

3O

 

 

 

 

60000 .
Defoliation I
I 51-100% _
50000 - 0 31-50% 84% \E
0 0-300%) I .-
1? I
E 40000 . :
El 2
.52 I
E 30000 - 2
3 ° -‘
a g _
$ 20000 - .0
III 0'
6563 EM/ha :
O ' 0
10000 - o \ I <2)-
....... o............. ....o.....o.*...l..
O O .
O O O
A Q) 0 O 8
o ' I l I l l
0 20 40 60 80 100

% new egg masses

Figure 4. Percentage of new gypsy moth egg masses versus egg mass density from the
fixed-radius plots. Defoliation was grouped into three categories. All stands with over
30% defoliation had at least 84% new EM and 6563 EM/ha, as indicated by the dotted
lines. n=32 sites.

31

 

 

 

 

60000 -
I Defoliation
. I 51-100°/o
500004 ; ; 0 31-50%
é. 0 0-30%
400ml
g E
a 30000 - I
a 0: °
5.3 E-
20000 - - o
I o :
10000- 0 I 0:0 .
...... O O---¢----------.- ..............
O .
o- 00 :68 I o
0 T l I OTC lo
10 15 20 25 30 35 40
Mean egg mass length (mm)

Figure 5. Gypsy moth egg mass (EM) length versus EM density in the fixed-radius
plots. Defoliation was grouped into three categories. All stands with over 30%
defoliation had mean EM lengths between 19.3 and 30.4 mm, and at least 6563 EM/ha,
as indicated by the dotted lines. n=31 sites.

32

 

 

 

 

40
Defoliation
35 - 0 31-50% _
0 O-30% ; 0
3 o
a 30 _n nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn ? IIIII on
E O .
' I
a: o o O
E" 25 o I oo
0 .o I
E o O: I
w I
an 20 - . .I
.. ............................. b0 , . . .0. .I.
O
O
15 - o
10 l 1 1 I '
0 20 40 60 80 100

% new egg masses

Figure 6. Percentage of new egg masses versus egg mass length. Defoliation was
grouped into three categories. All stands with over 30% defoliation had at least 84%
new EM and mean EM lengths between 19.3 and 30.4 mm, as indicated by the dotted
lines. n=31 sites.

33

 

 

 

 

 

 

 

 

 

 

60
Fixed-radius plots
50 ~L
E 40 4~
5
.E
5 30 4— . °
0
.§ . '0 ' '
E- 20 «~00 ‘
a: .‘ . .
'”I I I
10 “.
o . l l 1r l
0 10000 20000 30000 40000 50000 60000
Egg masses/ha
60
100-tree plots
I
50 ~~ 0
I
0 e
m 40 5*
0 I
E, 30 " ' °
0 '. a
E 0.
i" 20?
I
10 A».
0 4. l l l
0 2000 4000 6000 8000 10000
Egg masses/ha

Figure 7. Time required to conduct gypsy moth egg mass surveys using fixed-radius
plots and lOO-tree plots by egg mass density. n=32 sites.

34

 

 

 

 

 

 

100
90 ._ O E.ma1ma1ga
o NPV

80 4b
3 ..
E 70 "‘
“a 0
'3 60 ‘-
0.0
5
= ud-
§ 50 r
a.
0 40 "“- ‘1-
on
g
E 30 w 0

20 .- i

10 +-

0
0-30% 31-100%
Defoliation

Figure 8. Average percentage mortality caused by Entomophaga maimaiga and NPV
in gypsy moth cadavers collected at 28 study sites in mid-July, 1995 in the Huron-
Manistee National Forest, MI. Error bars indicate SEM. Average percentage of
mortality varied between the 2 defoliation levels for E. maimaiga (p=0.004) and NPV
(p=0.001). Percentage mortality varied between the 2 pathogens for both low
(p<0.001) and high (p=0.033) defoliation. Tests were conducted with the
nonparametric Wilcoxon two-sample test.

CHAPTER 2

Estimation of Seasonal Leaf Area Index with a Sunfleck Ceptometer

INTRODUCTION

Leaf area is frequently measured in forest stands because of its close association with
photosynthesis, transpiration, and forest productivity (McLeod and Running 1988). Leaf
area index (LAI) is a measurement of the one-sided leaf area per unit ground area (mzm'z).
Direct estimation of LAI using litterfall analysis or destructive sampling is time-consuming
and labor intensive (Bolstad and Gower 1990, Chason et al. 1991, Huston and Isebrands
1995, Vose et al. 1995). Allometric equations relating LAI to sapwood basal area have
been developed for many tree species. However, these equations require destructive
sampling and are often specific to the geographic area and year of development (Gazarini
et al. 1990, Burton et al. 1991, Huston and Isebrands 1995, Vose et al. 1995). The
amount of light transmitted through the canopy can be converted to LAI based on the
Beer-Lambert Law. This is an accurate and cost-efficient method of indirectly quantifying
LAI, especially when many stands must be sampled (Bolstad and Gower 1990).

Many studies have evaluated the Sunfleck Ceptometer® (Decagon Devices, Pullman,

WA) for use in measuring canopy transmittance (e. g. Pierce and Running 1988, Bolstad

35

36
and Gower 1990, Vose and Swank 1990, Vose et al. 1995). This device measures

photosynthetically active radiation (PAR) in the wavelength range of 400-700 nanometers
with a linear band of 80 sensors spaced at l centimeter intervals. It calculates the
arithmetic mean of values from each of the 80 sensors. Several readings can be taken and
then automatically averaged and stored in an internal data storage unit.

A potential use of the Sunfleck Ceptometer may be to quantify defoliation, such as
that resulting from herbivorous insect feeding. Theoretically, the amount of leaf area lost
to a defoliator could be quantified by measuring the increase in light transmittance through
a canopy. One of the major defoliating insect pests of hardwood forests in the
northeastern United States is the gypsy moth, Lymantria dispar L. (Lepidoptera:
Lymantriidae). Outbreak populations of this insect can defoliate millions of hectares of
forest in a single year (Butalla 1996). Defoliation is typically estimated visually and
assigned to qualitative classes such as low, moderate, and high (Talerico 1981). Observer
bias is fiequently a problem, however, and it may be difficult to consistently compare
qualitative estimates of defoliation (J obidon 1992).

We used a ceptometer to measure LAI periodically from initial leaf expansion to
autumn leaf drop in several hardwood stands representing a range of ages and species.
The objectives of this study were to examine seasonal within-stand variation in LAI and to
evaluate the potential use of a ceptometer to estimate gypsy moth defoliation. We also
examined potential associations between LAI and stand variables including basal area, tree

density, and average trunk diameter.

37

METHODS
Study site

This study was conducted at the Michigan State University W. K. Kellogg
Experimental Forest in Kalamazoo Co., Michigan (latitude, 42°21’29.0” to 42°22‘22.8”;
longitude, 85°20’50. 1” to 85°22’22.4”). The 240-ha forest consists of a variety of genetic
plantations and naturally regenerated hardwood stands. Nine hardwood stands that were
part of a related study on gypsy moth biocontrol (see Chap. 3) were chosen for this study
(Table 1). An experimental plot approximately 0.4 ha in size was established in each stand
in May 1994 and the plot center marked with a stake. Wire flags were placed 3 m apart
along transects radiating outward 30 m from the stake in the 4 cardinal directions (40 flags
per plot, Figure 1). The 9 stands were grouped in blocks of 3 based on their spatial
distribution within Kellogg Forest. Stand sampling order was randomized on each
sampling date, first by block and then by stand within block.

PAR readings were taken with a Sunfleck Ceptometer at approximately 1 - 2 wk
intervals from 11 June to 22 October 1994, and 29 April to 13 October 1995. Sampling
was done on clear (cloud-free) days, primarily between the hours of 1100 and 1430
EDST, although sometimes sampling had to start as early as 1000 and end as late as 1615.
Graphical inspection of the data indicated that time of day did not introduce a significant
bias to the LAI estimates. Each stand was sampled 4 - 7 times in 1994 and 9 - 10 times in
1995. Readings were taken at each flagged point with the ceptometer held horizontally 4
feet above the ground. The observer slowly rotated in a circle, taking a reading every 45

degrees. The average of the 8 values at each flag was stored in the ceptometer’s memory,

38
resulting in a total of 25,600 individual PAR readings per plot on each sampling date.

When clouds were present around the sun, sampling was halted until they had passed.
Data was downloaded to a computer at the end of each day.

A measure of above-canopy (or total incoming) PAR was taken in a nearby clearing
immediately before and alter taking readings within each stand. A diffuse PAR reading,
measuring reflected and scattered light, was obtained in a clearing by using the
ceptometer’s single sensor mode and shading the tip with an index card held
approximately 1 m above the sensor (Decagon Devices 1987).

Stand variables were measured in four 0.01-ha fixed-radius plots, 1 along each of the 4
transects within each experimental plot, with each centered 10 m from the plot center
(Figure 1). Basal area (cross-sectional area of tree trunks at breast height, or 1.37 m
above ground), tree diameter at breast height (DBH), density, and height were measured

for all trees at least 2.5 cm DBH, and means were calculated for each stand (Table 1).

Data analysis

No leaves were present on the trees during the initial sampling period in 1995 (29
April). These readings were used to estimate the woody area index (W Al), the tree bole
and branch portion of the total LAI in each stand. LAI was calculated using equations
based on the Beer-Lambert Law, where LAI is directly proportional to the logarithm of
the canopy transmittance (Q1/Qo) (Decagon Devices, 1987). Q, is the PAR reading under
the canopy, and Q0 is the above-canopy PAR reading. Consecutive sets of above-canopy
PAR and diffiise PAR readings were averaged and used in the LAI equations with

beneath-canopy PAR readings recorded during that time fi'ame. WAI was subtracted from

39

the LAI estimates of each stand to yield LAIs representing foliage only. Final readings
each year were taken after leaf abscission had begun.

Ceptometer transects in 5 of the 9 stands extended slightly beyond the edge of the
stand either into a clearing or a different stand. Some individual sampling points at the
end of transects were located outside the main stand because some plot centers were too
close to the stand edges. We dropped 1 - 10 points from the calculation of LAI in 5
stands because they fell outside of the intended stand. This was done consistently for all
sampling dates during both years of the study.

Associations among stand variables, WAI and LAI were evaluated using Pearson’s
product-moment correlation (r) (Sokal and Rohlf 1995). We computed the coefficient of
variation (CV) of LAI estimates for each plot for sampling periods between 11 June and 4
September for each year to examine within-season variability of the LAI estimates. This
was the period after fiill leaf expansion and before leaf fall, where LAI was expected to
remain relatively consistent. A t-test was used to determine if average stand LAIs varied
significantly between 1994 and 1995 for samples taken between these same dates. All
significance tests were conduted with a=0.05, using JMP Statistical Software (SAS

Institute 1995).

RESULTS

There were only 3 significant correlations among stand variables: basal area and stand
age (r=0.872, p=0.002), tree density and average DBH (r=-0.782, p=0.013), and average

height and average DBH (F0965, p<0.0001). All 3 associations were expected. Woody

40
area index (WAI) was correlated with stand basal area (r=0.841, p=0.005) but not with

age, average DBH, density, or height. Neither maximum LAI nor mean summer LAI
were significantly associated with stand variables or WAI consistently in both years.
Maximum stand LAIs in 1994 and 1995 were significantly correlated with each other
(r=0.741, p=0.022), as were average summer LAIs (r=0.773, p=0.015).

As expected, measured LAI increased quickly until early-June, remained relatively
stable during the summer and then began declining in early September (Figure 2). The
highest average summer LAIs (between 11 June and 4 Sept.) were in stands 40, 42 and 53
(Table 2). These were older stands (50 to 75 years old) with closed canopies dominated
by oak. The range of LAI estimates in this study (3.8 to 7.5) correspond well with those
reported in other hardwood stands (Bolstad and Gower 1990, Chason et al. 1991, Vose et
al. 1995).

Average summer LAIs differed between 1994 and 1995 in stands 15 (t= -2.46, df=6,
p=0.049), 16 (t= i-3.36, df=7, p=0.012) and 40 (t= 4.85, df=9, p=0.0009). There were no
significant differences in LAI between years for the other 6 stands. The 2 English oak
stands (15 and 16) had the lowest LAIs in 1994 (Tables 2 and 3). These stands also had
the greatest increase in LAI over the 2 yrs. Trees in these plantations averaged 17 it in
height, and grew vigorously, probably accounting for the significant increase in LAI from
1994 to 1995. LAI values were quite similar for the 2 stands, as expected for stands of
similar age, height and stocking density. The large decrease in LAI in stand 40 may have
occurred after several trees blew down during a Windstorm.

Coefficients of variation (CV) for stand LAI values during full leaf expansion ranged

from 6.9 to 23.5% (Table 2). Low CV indicated that LAI values estimated with the

41

ceptometer were consistent. CV were consistently low in stands 40 and 53 but there was
no consistent association among CV and stand variables over the 2 yrs.

The ceptometer provided a relatively quick way to measure PAR transmittance in
forested stands. Seven to 35 min were required to take the 40 transmittance readings in
each stand, with an average of 14 min (SEM=0.461, n=141) and a mode of 12 min (Figure
3). Average time decreased from 17 min in 1994 to 13 in 1995, showing that observer
experience speeded up the sampling. Many of the high time values occurred when the
observer waited for clouds to move away from the sun. Clouds were common in this area

due to the effects of nearby Lake Michigan.

DISCUSSION

Results from other studies that have evaluated the association of LAI with stand
attributes such as basal area, stand density, average diameter, and tree height have varied.
Liebhold et al. (1988) found that oak LAI measured from canopy photographs was
significantly correlated with basal area measurements. Gresham (1982) and Dalla-Tea and
Jokela (1991) noted an association between litterfall and basal area in pine stands. But
Vose et al. (1995) found no clear associations between litterfall and several stand variables
in hardwood stands, possibly because of the small range of the variables in their study.

The primary reason for the lack of significant associations in our study may be that species
composition varied considerably among sites. Associations among LAI and other stand
attributes vary with canopy structure. Our stands were composed of both pure and mixed

hardwoods species across a wide range of age, diameter, density, and height.

42

Accuracy of LAI estimates acquired using the Sunfleck Ceptometer have been
validated by more direct methods of LAI estimation like litterfall collection (Maass et al.
1995) and allometric equations developed from destructive sampling (Pierce and Running
1988, Bolstad and Gower 1990, Vose and Swank 1990). Clinton (1995) used a
ceptometer to find significant differences in PAR levels in hardwood forests with and
without Rhododendron understories. Jobidon (1992) successfiilly used a ceptometer to
measure PAR that was available to young spruce as a means of measuring efficacy of
herbicides used to control competing vegetation. The nationwide Forest Health
Monitoring (FHM) program, a joint Environmental Protection Agency/US. Forest
Service project, currently uses ceptometers to estimate forest LAI as a means of
evaluating the long-term health of the country’s forests (Tallent-Halsell 1994, Huston and
Isebrands 1995).

We found that the Sunfleck Ceptometer was a cost-efficient method of determining
LAI. Since it measures transmittance with 80 sensors at once, it was possible to collect
25,600 point samples in each plot in an average of only 14 min. This compares well with
Fassnacht et al. (1994), who estimated LAI using 3 light-measuring instruments. They
found that the LAI-2000 Plant Canopy Analyzer (LI-COR, Lincoln, NE) took only 4 min
per plot, the Sunfleck Ceptometer required 22 min, and the DEMON (C SIRO, Canberra,
Australia) required 4 hrs.

Leaf area index in a forest stand can vary from year to year due to local site
conditions, weather variability between years, tree death, and plant resource allocation
patterns such as periodic production of large seed crops (Burton et al. 1991, Huston and

Isebrands 1995). Burton et al. (1991) observed a 20 to 30% reduction in LAI during the

43

year of a prolific seed crop in a northern hardwood stand dominated by sugar maple (A oer
saccharum Marsh.) We observed significant between-year differences in average summer
LAI in 3 of our 9 stands. Two stands were young and vigorously growing plantations that
had not yet reached their maximum LAI potential. The loss of several trees that blew
down in a Windstorm likely caused the decreased LAI in the other stand.

Ceptometers have been found to be sensitive to within-season changes in PAR and
LAI (Vose and Swank 1990, Maass et a1. 1995) and provide consistent results (Bolstad
and Gower 1990). In our study, the method was sensitive to foliage production for
measurements taken during the spring of 1995 (Figure 2). LA] estimates during the
summers of 1994 and 1995 were relatively consistent, with an average coefficient of
variation of 11.5. Temporal variation within the summer may have been the result of
natural variation in the method or from weather conditions. Movement of leaves and
branches by the wind can cause variation in the LAI estimated by a ceptometer (Péch
1986). Variation in summer LAI was greatest in the yellow birch stand (#23, Figure 2,
Table 2) in 1995. This stand was thinned in October 1994, so the trees may have been
adjusting to the increased amount of sunlight and ispace. Another possible explanation is
unnoticed growth and shedding of reproductive catkins on the yellow birch trees.

We believe that the ceptometer could be a cost-efficient method of estimating
defoliation. Liebhold et al. (1988) estimated gypsy moth defoliation in oak by evaluating
the change in LAI estimated from canopy photographs. Burton et al. (1991) detected
significant differences in LAI between northern hardwood sites using a ceptometer. In
that study, reductions in LAI were only 20 to 30%, caused in one instance by a defoliating

insect, and in another instance by a prolific seed crop.

44

Due to the inherent variability of the method, a single ceptometer plot per stand, as in
our study, may only allow the user to classify defoliation into 3 or 4 broad categories. But
the ceptometer will provide a method of measuring defoliation that is repeatable,
quantitative, and free from observer bias (Jobidon 1992). Low levels of defoliation may
be difficult to detect with a ceptometer. Although using multiple plots may increase the
accuracy and precision of the estimates, it may counterbalance the advantage of its low
time requirement. It may also be difiicult to account for the spatial variation of
defoliation, especially in large forest tracts. Sampling along a transect through the stand
may give a more representative estimate of defoliation than clumped samples within small
plots.

Many studies estimating forest LAI have used several light sensors simultaneously
(Chason et al. 1991, Tallent-Halsell 1994). A remote unit measures total incoming PAR
in a clearing, while another unit, like a ceptometer, is operated manually below the
canopy. Readings are then paired according to the times they were recorded. If resources
permit, use of multiple sensors simultaneously should improve accuracy and precision of
LAI measurements.

In the case of gypsy moth and late-season defoliators, which feed primarily after full
leaf expansion, the before- and after-defoliation readings could be taken in the same year.
Additional measurements could be taken to estimate refoliation after major defoliation
events. Timing of measurements depend on local tree and defoliator phenology. Liebhold
et al. (1988) found that oak LAI reached a plateau by 1 June in Massachusetts; leaf
expansion at our site in southern Michigan was finished by the first week of June. For

early-season defoliators that cause significant leaf damage before full leaf expansion, an

45

initial LAI estimate should be taken during the year prior to defoliation. This may not
work in stands where LAI may vary greatly between the 2 yrs, like young rapidly-growing
stands or stands with significant tree mortality or windthrow. Insect population
monitoring with pheromone traps, egg mass surveys, etc. may be needed to predict where
defoliation will occur. Using ceptometers to provide estimates of leaf area lost to
defoliators could help in predicting impacts on nutrient movement in forest ecosystems
(Mattson and Addy 1975, Swank et al. 1981, Grace 1986). Measuring the increased
amount of light coming through a defoliated canopy may also be useful in examining
changes in understory regeneration or impacts on ground-dwelling arthropods and other

wildlife sensitive to such changes.

 

46

3 _ 8: E: m. _ m :2 3532 8382 .928 8.83 3.8 :85 mm

 

o4 _ mm: 2: wsm $2 3532 292: v8 Jae com NV
m.m_ 32 Num— vsm mma confine—m uni 823 68333 i8 3M ow
m. _ m Sc 0.2 @dm ~02 confine—m xwo com mm
m.m_ N. _ 2 NC v. 8 $2 .538qu :85 3o=o> mm
5: N2: 3 _ . 3m $2 3532 means .292: we .83 58sz mm
o._ _ .2 _ _ o.m_ 9mm $3 5:853 .92? x33 Jae—m3 xoflm a
mam me: _.o m.m 23— 83355 . x8 aim—mam @—
N.m 32 mo m.m 33— 538983 xwo smzmcm m—
AEV 33:5me A83 35.5 Ewto 98?. 3.58: 860% bombs :85an .o: wcfim

Ems: bacon Ema .w>< 38 325 28m 8 5:855

 

.zmo :8 Wm M moot =a 8m 82 5 3.532: 203 Emma: new $3ch .mmn ammonia .88 Baum dz
coo 8882mm 5 «monon— wwofiv. .v. .3 2: an mncflm 295m 05: 05 “8 83388.85 “Edam .«o 35:26 ._ 29d.

 

47

2: ed as- :3 3.: 3:888:88 x8338 scam mm

 

 

3 3 _ $5. a; o: v.8 we .288 B: N...
3 E Q. _ - 2.: m2 Beams 65: 323 .818 B: 9.
3: 3: N3- Re ”is x8 8: mm
mm ad 2:. as 8.: :25 26:“; mm
:2 3 3o- 8: 03 23322 .292: B: :83 58:3: 8
a: 4.x 3.? m: o3. 92: x83 .55? 8.8.: a
2; 4.: 91+ w _ .m w: x8 imam 2
ed 3; 8. _+ N3 2:. :8 imam 2.
3% gov A~-E~Ev A~-E~Ev ANENEV 88on beams; Eafifiom .o: :Sam
82 82 32 2 82 82 5863 .32 58:3
>o 24 see a 8:20 :3 :82

 

5:0 owmza 8:22: 24 .30» some .Eom v 9.8 0:3. I 5253 32:3 wEEEfl :8 m??—
ngm wfims @3358 083 >0 :5 F5 owfio>< .=2 :00 0:52:23. :m wagon wwozom .M .3 05 E 35% 29:8
vi: 2: :8 56:85 («:2 :3 a: 98 :ozatg mo 330508 :5 A25 :35 :8: .32 038% MO bus—Ea .m 033.

 

48

3:. m3 :2. 2: 3:33... $56 8.83 :8 :22: mm

 

 

m _ . _- S: m _ .w R: x8 :2 .288 B: a.
S. _- on: S: a; Beans as: 223 :8 B: 2.
NS: 43. NS. :3 x8 8: mm
NS- $8 5.: So . :25 25:3, mm
:3. 3.: 3.: w _ ._ 3:33... .292. we :83 58me mm
03+ :3 $4. 3o 92: :83 .35? 82: a
a. _+ a: :3 as 8.8 5:8: 2
a. _+ :3 :1. one :8 5:2: 2
AGENEV A~-E~Ev A~-E~EV A~-E~Ev $6on @9230 5:5an .o: 98%
32 9 as: 82 $2
:5 gas 5 8:52: :3 5:832 23

 

5:0 mo>:2 8:205 73 8:82:82 22 .60 coughs—:2 5 395m wwo=o¥ 2 .3 2:
:: 25:5 29:3 2:: 2: 89¢ 3:1: 5:5 :2: («:2 89:38:: 2:: C<3V x262 :8: .383 .«o 3:995 .m 2:3.

49

xkxxkxxxxx

xxxxxxixxfix xixxxzxxxxxx

l-—— 33m ———1

xxxxxxxxa’x

Figure 1. Plot layout for ceptometer study. Each ‘x’ marks a sampling point. Stand
attributes (basal area, diameter, height, and density) were measured in four 0.01 ha plots
indicated by the circles.

{gm 5:: :0 8:8: to. :25:

:5 E 8: $383: 5:85:53 33m :53 3.6365 3 32 n:: woo. m:::: 3:: 0926.: :35 83.33 .N Eswfi

 

  
 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8.5 SEE. 8:: SE3. 3: 5:2.
com :2 SN :2 2: can on: can :2 :2 8m :2 45m :2 2:
w :11 : 1 w w w a o w 1. w - o
1 Nu.
m:
1 m
n
1 m
can
1 o
m. N n
m.
B
s .1...
a
s p
n
211 w 211 w :11 w
8: :2 8a :2 2: 8m c2 :8 o: 2: 2:
1 o
1 NW.
n
I vm
B
y OM-
n
32......32lol 211: §T.i:32|¢| E11: §_..§..§_|¢| my:

 

 

 

 

 

 

 

 

 

51

 

 

28-30 31-33 34-36

19-21 22-24 25-27

16-18

5

2.. a
. 3.3.) 1.5%.

       

         

.
3.1.1.2....17 .I..f..u...;-...-..If.-..¢..I
. . . u

- . r .1 31%|. (clot-II- !I‘li nnnnnnnnnnnnnnnnn
.z.....vv......r$rv$.... ”N5. .1 viafiavvzv.)

 

 

 

50<»
4O «-

35:99.5

Time (minutes)

ired for field sampling of one plot (40

me requ

ion of t'

but

ceptometer readings) in 1994 and 1995 (n=141).

istri

Figure 3. Frequency d

CHAPTER 3

Evaluation of Gypsy Moth Biological Control at the W. K. Kellogg Experimental

Forest, Michigan, Including Introductions of Two Natural Enemies

INTRODUCTION

Introductions of exotic parasitoids, predators, and pathogens of the gypsy moth have
taken place in the United States intermittently since 1905 (Hoy 1976). By 1980, at least
11 exotic natural enemies were well-established in the US. (Barbosa et al. 1975, Leonard
1981). Two exotic natural enemies that have become established are the braconid wasp,
C otesia melanoscela (Ratzeburg), and the fimgal pathogen, Entomophaga maimaiga
Humber, Shimazu and Soper. Cotesia melanoscela was first introduced in the eastern
US. in 1911 (W eseloh 1981). This bivoltine wasp parasitizes young gypsy moth larvae,
preferring first and second instars (W eseloh 197 6). C. melanoscela is easily reared in the
laboratory, and can be obtained from commercial suppliers for release against gypsy moth
(D. McCullough, pers. comm).

The firngus E. maimaiga was first brought to the US. from its native Japan in 1910
and released in Massachusetts. No evidence of establishment was found and the project
was deemed a failure (Hajek et al. 1995). It mysteriously appeared in 1989 when an

epizootic of the fiingus swept through gypsy moth populations across 7 Northeastern

52

53
States (Andreadis and Weseloh 1990, Hajek et al. 1990a). The fiingus is dependent on

weather, needing wet, humid conditions for spore germination and gypsy moth infection
(Hajek et al. 1990b, 1993). Entomophaga maimaiga is not commercially available at this
time.

The objective of this study was to introduce C. melanoscela and E. maimaiga in a
low-level gypsy moth population along the leading edge of the gypsy moth advance, and
observe whether these natural enemies became established or affected the gypsy moth
population in any measurable way. Our second objective was to survey for other natural
enemies that were present in this gypsy moth population to determine if exotics had
moved into the area with gypsy moth. Our third objective was to determine what native
natural enemies were parasitizing or preying on gypsy moth larvae. This baseline
information will be usefiil for gypsy moth managers who must evaluate releases of
biological control agents to avoid overlap of ecological niches between biocontrol releases

and natural enemies already present.

METHODS

This study was conducted at the Michigan State University W. K. Kellogg
Experimental Forest in Kalamazoo County in southwestern Michigan. The 240-ha forest
consists of a variety of genetic plantations and naturally regenerated stands of many
hardwood and conifer tree species. This area was selected because low-level gypsy moth
p0pulations had been present for several years, but no significant defoliation had yet

occurred.

54
A total of 68 gypsy moth pheromone traps were placed along a grid at a density of 28

traps per km2 throughout Kellogg Forest between 1993 and 1996. Egg mass surveys were
conducted each year afier leaf fall around the 33 trap locations that were within hardwood
stands. Four 0.01 ha plots, with plot centers located 10 m from the pheromone trap
location in the cardinal directions, were used to determine mean egg mass density
(Kolodny-Hirsch 1986).

The 9 stands with the highest egg mass counts in 1993 were chosen as release sites for
E. maimaiga and C. melanoscela in 1994 and E. maimaiga only in 1995 (Table 2). The 9
stands were grouped in blocks of 3 based on their spatial distribution within the forest.
The 3 treatments of E. maimaiga, C. melanoscela, and control were randomly assigned to
the 3 stands in each block (Table 1). Basal area (cross-sectional area of tree trunks at
breast height, or 1.37 m above ground), tree diameter at breast height (DBH), density, and
height were measured for all trees with at least 2.5 cm DBH within the 0.01 ha plots used
for egg mass surveys (Table 1). Egg masses within plots were monitored to determine

hatching date and to monitor larval development.

Biological control introductions

Permits for shipment and release of the biological control agents were obtained from
appropriate state and federal agencies. C. melanoscela were purchased from National
Gypsy Moth Management Group, Inc., Landisburg, PA, and wasps were shipped as
cocoons in cardboard cartons on 10 May 1994. We placed approximately 5000 Cotesia
cocoons at the center of each of 3 release plots on 12 May (see Table 1). Paper cartons
containing 500 cocoons each were stapled to tree boles facing away from the sun. C otesia

adult emergence and dispersal was completed within 4 d. Gypsy moths were first and

55

second instars during the release period. The cartons of cocoons were collected from the
field after adult emergence was complete so that we could determine emergence rate. A
total of 100 cocoons from 8 randomly selected cartons were examined. Also, a carton of
500 cocoons was retained so we could determine the ratio of males to females that
emerged (adults were allowed to eclose but were not released). No C. melanoscela were
released in 1995 so that we could continue to evaluate the 1994 introduction.

We introduced E. maimaiga to the W. K. Kellogg Forest in 1994 and 1995 via soil
containing resting spores. Soil for the 1994 introduction was collected by Frank Sapio
(Division of Forestry, Michigan Department of Natural Resources) from Lake County in
northwest Lower Michigan, and was tested for plant parasitic nematodes before
introduction. This soil was acquired from one of the original release sites where Smitley
et al. (1995) first introduced the pathogen into the state in 1991. An E. maimaiga
epizootic occurred at this site in 1993 (Smitley et al. 1995). Soil for the 1995 introduction
was collected by Dr. Joseph Elkinton (Department of Entomology, University of
Massachusetts) at Holyoke Range State Park near Amherst, Massachusetts, where the
gypsy moth population had an epizootic of E. maimaiga in 1994. Soil for both releases
was collected from the bases of oak trees. Samples of the soil from the 2 introductions
were sent to Dr. Ann Hajek (Department of Entomology, Cornell University) for
quantification of resting spores concentration. Additional soil samples were saved for
later use in bioassays with gypsy moth larvae.

Six trees within 11 m of the plot center were chosen in each of the 3 E. maimaiga
release plots at Kellogg Forest (Table 1). Preference was given to large trees and to

species preferred by the gypsy moth. Leaf litter was scraped away from the base of

56

release trees, exposing bare soil. One liter of soil was spread around the base of each
release tree on 12 May 1994 and 22 May 1995, and leaf litter was replaced. The soil
around the release trees was watered once or twice weekly (depending on rainfall) from
the introduction dates until 18 June 1994 and 23 June 1995 to provide moisture for resting
spore germination (Hajek and Roberts 1991). Two to 3 liters of water were used around

each release tree.

Entomophaga maimaiga bioassay - 1995

We started a bioassay on 23 May 1995 to determine whether the soil obtained from
Massachusetts had adequate viable resting spores to cause infection in gypsy moth larvae.
A thin layer of soil was spread over the bottom of a single clear plastic container (25 x 19
x 9 cm). Gypsy moth egg masses were obtained from the rearing facilities at the USDA
APHIS rearing facility at the Otis Air Force Base in Massachusetts. Larvae were reared
on artificial diet (Bell et al. 1981) until they were third instars. Thirty third-instars were
placed on the soil along with cubes of artificial diet. The closed container was placed in a
growth chamber at 18 to 20°C and a photoperiod of 15:9 (LzD). Soil was kept moist but
not wet. Fifteen larvae were removed from the container on 27 May and placed in
individual cups with artificial diet but no soil. The remaining 15 larvae were left with the
soil until 29 May, when they were removed and placed into individual diet cups. Cadavers

were examined to determine if pathogens were present.

Entomophaga maimaiga bioassay - 1996

We conducted a second bioassay for E. maimaiga in 1996 using soil samples from the

W. K. Kellogg Forest, the Huron-Manistee National Forest, and lefiover soil from the

57

1994 and 1995 releases. Soil from around 6 trees was collected and pooled for each one-
half liter sample from the W. K. Kellogg Forest and the Huron-Manistee National Forest.
Soil was scraped to a depth of 1.5 cm within 25 cm of the base of each tree. Sixteen soil
samples were collected at the W. K. Kellogg Forest on 3 May 1996 from the 9 treatment
plots and from 4 additional non-release sites scattered throughout the forest. One sample
was collected in each control and C otesia release plot. Two samples were collected
within each E. maimaiga release plot, one from around the trees where the fungus was
released in 1994 and 1995, and one from around non-release trees within 10 m of release
trees. This was done because resting spores from the actual releases were likely present in
the soil around release trees, whereas any spores present around non-release trees would
likely be from infected larvae in the plots.

On 7 May 1996 we collected soil from 6 sites within the Huron-Manistee National
Forest in northwestern Michigan. These samples were intended to be positive controls for
the bioassay, since cadavers infected with E. maimaiga were found at these sites after
peak 1995 defoliation. Spring 1995 egg mass densities ranged from 790 to 8814 egg
masses per ha, and defoliation was low (0 to 20%). in these sites. We also tested soil that
had been saved from the 2 releases at Kellogg Forest (2 samples from each).
Approximately 50 g of soil (wet weight) fi'om each of the 26 samples were placed in large
petri dishes (25 x 150 mm). Ten third-instar gypsy moth larvae (eggs obtained from Otis
Air Force Base) were put on each soil sample on 14 and 15 May 1996. The petri dishes
were maintained at room temperature (20 - 25°C) and the soil was kept moist. Larvae
were transferred from the soil to individual diet cups (without soil) on 21 and 22 May.

Cadavers were examined for the presence of pathogens.

58

Subsamples of 9 soil samples collected at Kellogg Forest for the 1996 bioassay were
also sent to Dr. Ann Hajek for quantification of resting spore density (Hajek and Wheeler
1994). These included soil samples from the 3 control plots and the 3 E. maimaiga
release plots (release trees and non-release trees). Each of the 9 subsamples was analyzed

separately.

Larval sampling

Since gypsy moth populations were low, we used burlap bands to sample larvae in
1994 to 1996 (Reardon 1976). Bands were wrapped around trees along transects in the 4
cardinal directions to a distance of 30 m from the plot center. We banded 18 trees in each
plot (2 near the center, and 4 along each transect) at a height of 1.3 m above ground.

Bands were checked every 1 to 2 wk during the period from early instar to pupation.
In 1994 and 1995, healthy-looking larvae and pupae were counted (Figures 1 - 3), while
those that appeared dead or unhealthy were collected in individual containers. All larvae
under bands were collected in 1996. We also collected potential predators of the gypsy
moth under the burlap and noted interactions between gypsy moths and predators on trees

without burlap bands.

Mortality diagnosis

Most of the gypsy moth larvae collected from burlap bands were dead when collected.
Live larvae were reared on artificial diet until they pupated, died, or a parasitoid emerged.
These cadavers and those from the bioassays were stored at room temperature for 2 d
after death to give any internal pathogens time to develop to make diagnosis easier.

Cadavers were then fi'ozen until they could be examined. For pathogen diagnosis, a

59

microscope slide was made of tissue from the cadaver. A drop of lactophenol cotton blue
firngal stain was added to the slide (Poinar and Thomas 1978, Smitley et al. 1995).
Diagnosis was made with a phase-contrast compound microscope, using lOOX for E.
maimaiga conidia and resting spores, and 400x for NPV polyhedral inclusion bodies

(PIBs) (Andreadis and Weseloh 1990, Smitley et al. 1995).

Statistical analyses

Our data included annual gypsy moth pheromone trap counts (1993 to 1996, up to 68
locations), annual egg mass counts (1993 to 1995, 33 locations), and annual larval counts
(1994 to 1996 from the 9 treatment plots). We used correlation coefficients to examine
associations among the pheromone trap, egg mass and larval counts. Pearson’s product
moment correlation (r) (Sokal and Rohlf 1995) was used to analyze the majority of the
data. Logarithmic and square root transformations were used when appropriate to meet
the assumption of normality. Correlations using pheromone trap counts. only (between
years) used data from the 68 traps across Kellogg Forest. Correlations of pheromone trap
counts with egg mass counts used data from the 33 sites where both types of data were
collected; Spearman’s nonparametric rank correlation (r.) (Sokal and Rohlf 1995) was
used because of the high frequency of zeros for egg mass counts. We only used data from
the 9 treatment plots for product-moment correlations involving larval counts.

Analysis of variance (ANOVA) was used to test for differences in mean E. maimaiga
resting spore densities among the 1996 soil samples tested by Dr. Hajek. All statistical
analyses were conducted with IMP Statistical Software (SAS Institute 1995) using a

significance level of alpha = 0.05.

60

RESULTS

Contrary to expectations, gypsy moth population levels remained low during the 3 yrs
of this study. Pheromone trap counts and egg mass densities are summarized in Table 2.
We conducted annual defoliation surveys in the study plots. Although we observed gypsy
moth feeding in many plots, defoliation never exceeded 1% during this study.

Gypsy moth larval densities in the treatment plots were roughly associated with
abundance of preferred hosts (Figures 1 - 3, Table 1). Plot 16, an English oak plantation
(a preferred host), had the greatest number of larvae. A maximum of 103 larvae were
observed under the 18 burlap bands during a single sampling period. We found only 3
larvae in the yellow birch stand (plot 23) during the 3 yrs of sampling. Yellow birch is not
a preferred host for gypsy moth (Liebhold et al. 1995). Other stands with high larval
counts had high proportions of oak species.

Pheromone trap counts from consecutive years were significantly correlated (r=0.798,
p<0.0001 for 1993 and 1994; r=0.402, p=0.001 for 1994 and 1995; r=0.422, p=0.018 for
1995 and 1996). Similar trends were observed for egg mass counts from consecutive
years (r,=0.583,p<0.001 for 1993 and 1994; r,=0.901,p<0.001 for 1994 and 1995).
Correlations between pheromone trap counts and counts of egg masses laid in the same
year were significant for 1993 and 1994 (r,=0.564, p=0.0006 and r,=0.645, p<0.001,
respectively), but not for 1995. Pheromone trap counts from 1994 were significantly
correlated with counts of egg masses laid in the previous year (r,=0.438, p=0.011), but

this association did not hold true for correlations of 1995 and 1996 trap counts with

61

preseason egg mass density. Counts of total larvae under burlap bands in each treatment
plot were not significantly associated with pheromone trap counts or with egg mass counts
in those plots. Larval counts from 1994 and 1995 were significantly correlated with each
other (r,=0.689, p=0.040), but correlations between larval counts in 1995 and 1996 were

not significant (r,=0.653, p=0.057).

Parasitoids, predators and pathogens

It is difficult to accurately determine parasitism rates in low density gypsy moth
populations since few larvae can be collected. Therefore, these results should be
interpreted with caution. Since we collected only the larvae that appeared to be dead or
dying, parasitism rates presented here may underestimate actual rates if some parasitized
larvae under the bands appeared healthy or if parasitized larvae remained in the tree
canopies.

We counted 2,122 gypsy moth larvae under burlap bands at Kellogg Forest during this
study. A total of 453 gypsy moth larvae were collected under the bands, with 194, 92,
and 167 in the years 1994, 1995, and 1996, respectively. Larvae collected in 1994 and
1995 were examined for parasitoids and pathogens, but larvae from 1996 were only
examined for pathogens. Of the 286 gypsy moth larvae that were collected in 1994 and
1995, 128 were killed by parasitoids, predators, or pathogens. Another 129 died from
unknown causes, and 29 were healthy. The largest individual source of larval mortality
was the tachinid fly Compsilura concinnata (Meigen). This parasitoid was reared from 36
larvae in 1994 and in 19 larvae in 1995. Its highest parasitism rates during a sampling
period were 27.3% on 28 June 1995 in plot 40 (3 of 11 gypsy moth larvae seen), and

13.4% on 30 June 1994 in plot 25 (11 of 82 larvae seen). Other tachinid flies collected

62

infrequently from gypsy moth larvae were Parasetigena Silvestris (Robineau-Desvoidy) (7
specimens), Eusisyropa virilis (Aldrich and Webber) (2 specimens), and Blepharipa
pratensis (Meigen) (1 specimen). Parasetigena silvestris also emerged from 4 gypsy
moth pupa collected under burlap bands. The encyrtid wasp Ooencyrtus kuvanae
(Howard) was observed each year parasitizing gypsy moth egg masses in many parts of
the forest. Phorid fly larvae were commonly found in dead late instar larvae and pupae.
Phorids are primarily scavengers, and Sabrosky and Reardon (1976) considered them to
be of no importance.

NPV was found in 26 larvae in 1994 and in 15 larvae in 1995. Polyhedral inclusion
bodies (PIBs) were present in small numbers in most of these larvae, but were abundant
and presumably responsible for mortality in only 5 larvae.

Many arthropods were observed preying on gypsy moth life stages. Stink bugs
(Hemiptera: Pentatomidae) were the predator most often seen attacking gypsy moth on
tree holes and under burlap bands. Both immatures and adults were active predators, and
were observed feeding on all gypsy moth life stages except eggs. Apateticus cynicus (Say)
was the predaceous species that we found most often, followed by Podisus modestus
(Dallas). Other pentatomids were P. maculiventris (Say), P. placidus Uhler, P.
serieventris Uhler, and Dendrocoris humeralis (Uhler). All of these species are generalist
predators, and all have previously been reported to prey on gypsy moth except P.
maculiventris, which is known to have a very large host range (McPherson 1982). All of
these pentatomid species are active during the period of the year when gypsy moth larvae

are present (McPherson 1982), and we found all 6 species under burlap bands.

63

Ants and spiders were only rarely observed feeding on gypsy moth larvae. Many other
predators were collected under burlap bands, but were not observed preying on gypsy
moth. These included spiders, harvestmen, and the carabids Calosoma scrutator Fabricius
and C. frigidum Kirby. These predators are opportunistic, feeding on whatever prey or

carrion they encounter (Smith and Lautenschlager 1978, Cameron and Reeves 1990).

Cotesia melanoscela

Emergence rate of C. melanoscela adults at the time of release was 47%, based on our
examination of 100 cocoons. Females comprised 54% of the adults that emerged (50
adults examined). Recovery of C. melanoscela in the field was low. All of the Cotesia we
recovered had already emerged from their hosts under burlap bands or on tree trunks or
foliage; none emerged from the gypsy moth larvae we collected. At least 38 cocoons
were found in treatment plots in 1994. The actual number may be as high as 55, because
cocoons could not always be collected and therefore may have been counted during more
than 1 sampling period. Thirteen cocoons were seen on tree boles and leaves, while the
rest were found under burlap bands. Nine to 15 cocoons were found in plot 15, 27 to 38
in plot 16, 1 in plot 25, and 1 in plot 53 (see Table 1 for plot descriptions). It is not
surprising that C otesia were found in 2 non-release plots (plots 16 and 25); they were
within 225 m of a release plot and both plots had a greater number of gypsy moth larvae
than the release plots. Only 3 cocoons were found in 1995, 1 each from plots 15, 16 and

53.

64

Entomophaga maimaiga

None of the larvae collected in 1994 and 1995 were infected with E. maimaiga. Six
infected larvae were collected in 1996, but only 1 was within a treatment plot (plot 15).
Four were collected under burlap bands on a small group of trees separated from the
nearest E. maimaiga release plot by about 100 m of grass. One infected larva was
collected on a tree bole near plot 53. All larvae infected with the fungus were collected on
8, 10 or 15 July.

Quantification of resting spores in soil. Resting spore densities in the soil samples
were determined by Dr. Ann Hajek’s laboratory according to the methods described in
Hajek and Wheeler (1994). Soil for the 1994 introduction contained 62.3 i: 13.1 (mean 5:
SE) resting spores/g of dry soil. Soil for the 1995 introductions contained 1676 :t 186
resting spores/ g.

Low numbers of resting spores were found in all 9 samples of soil collected in May,
1996 at the W. K. Kellogg Forest. The highest counts (22.2 i 8.7 resting spores/g) were
in soil from around the trees where resting spores were introduced in the first 2 yrs of this
study (18 trees total in the 3 release plots). Soil fi'om the 3 control plots, however, had
similar counts (20.1 3: 7.0 spores/g). Lowest counts (9.5 i 4.2 spores/g) also came from
E. maimaiga release plots, but from around trees where no resting spores had been
intentionally introduced. There were no significant differences in mean resting spore
density among samples from the 3 categories (F =0.959, df=26, p=0.4). Limited resources

prevented us from quantifying soil resting spore density in the Cotesia release plots.

65

Bioassays. In the bioassay that used soil from the 1995 E. maimaiga release, 7 larvae
were dead and producing E. maimaiga conidia as early as 29 May, 6 d after being placed
on the soil. Of the first 15 larvae removed from the soil, 10 were infected with E.
maimaiga and died, and 5 were successfully reared to adults. Of the second group, 9
larvae were killed by E. maimaiga, 3 survived to the pupal or adult stage, and 3 were
killed by NPV, presumably contracted from the soil. All larvae that were infected with the
fungus died by 4 June.

All of the larvae in the 1996 bioassay died except for 3 that pupated. Slides were
made of 146 of the 260 specimens. NPV was found in 34 of the larvae, but E. maimaiga
was not observed in any of the larvae. When NPV was present in the larvae, only small
numbers of PIBs were present in larvae reared on soil from Kellogg Forest. Large
numbers of PIBs were present in many of the larvae reared on soil from the Huron-
Manistee National Forest and on soil from the 1994 and 1995 firngus releases. Larvae
reared on these soil samples often had very high numbers of P183. NPV survived in soil
that was saved from the 1994 introduction, even after being frozen at -15°C for 2 years.
The Massachusetts soil (1995 release) was stored in a refrigerator (4°C) for the 1 1.5
months between the time when we obtained it and when we conducted the bioassay.

Since no E. maimaiga infection was noted in the larvae used in the bioassay, it appeared
that resting spores did not germinate in any soil sample, even in the samples where we
knew they were present. They may not have received the appropriate amount of moisture
while in the petri dishes, or may not have had adequate time to break diapause. The cause

of death could not be determined for 112 of the larvae using our evaluation techniques.

66

High larval mortality in the 1996 bioassay may have been caused by improperly prepared

artificial diet in the individual cups.

DISCUSSION

Introduction and establishment of C. melanoscela and E. maimaiga proved to be
difficult in the low-density gypsy moth populations at the W. K. Kellogg Forest. By the
end of the 3-yr study, it was still not clear how well they were established. The number of
C. melanoscela recovered declined afier the release, and E. maimaiga was first recovered
in 1996, although in very low numbers.

The inability of C. melanoscela to impact gypsy moth populations is usually attributed
to 2 factors, hyperparasites and poor host synchronization (Grimble and Palm 1976,
Weseloh 1976). Second-generation cocoons are vulnerable to attack from over 30 native
hyperparasite species from July until the following May (Van Sickle and Weseloh 1974).
Weseloh (1983) found the survival rate of overwintering cocoons in Connecticut to be
only 2.3%. High mortality resulted from hyperparasitism (66.4%), predation (23.4%) and
unknown causes (7.9%). Even with the low number of cocoons recovered at Kellogg
Forest, 2 were found to be hyperparasitized. We have also seen frequent hyperparasitism
of C. melanoscela in the Huron-Manistee National Forest (northwest Lower Michigan),
where Colesia is often quite abundant.

Cotesia melanoscela adults emerge from their overwintering cocoons when the gypsy

moth are first and second instars, and are quite successful at attacking these larvae. But

‘/_.

67

development of the second generation is not completed until most gypsy moth larvae are
fourth instars (Weseloh 1976). Successful attack rates decrease dramatically with fourth
instars because these larvae have long hairs and can shake off attacking wasps (Weseloh
1976, 1981). Wasps of the second generation started emerging from their cocoons around
7 June at Kellogg Forest. At this time gypsy moth larvae were primarily in the third and
fourth instars. C otesia melanoscela appeared, therefore, to have the same problems at
Kellogg Forest that it has in other parts of its range.

The lack of success of E. maimaiga in 1994 and 1995 is difficult to explain. Soil from
the 1994 introduction had a very low concentration of resting spores (62.3 i 13.1
spores/ g), but the inoculum in 1995 had high numbers of resting spores (1676 :L- 186
spores/ g) that were shown to be viable through a bioassay. Previous introductions in
northern Michigan were made with soil that had counts of 937 spores/ g (Smitley et al.
1995). Hajek and Roberts ( 1991) successfirlly introduced E. maimaiga in Ithaca, New
York using soil with 320 resting spores/g. Given this data, soil used in the 1995
introduction at Kellogg Forest had an ample amount of resting spores to cause infections.

During the 1989 outbreak of E. maimaiga in New England, high rates of mortality
were only observed where total May and June rainfall totaled at least 20 cm (Elkinton et
al. 1991). May and June rainfall at Kellogg Forest totaled less than 19 cm in 1994 and
1995, and 22 cm in 1996 (Table 3). Rainfall during the first 2 yrs of the study may have
been too low to result in successfiil infection of larvae. Although we periodically watered
the area where soil was released between precipitation events, the soil ofien remained dry
for 2 or 3 consecutive days. In 1996, infected larvae were found at Kellogg Forest

between 8 and 15 July. Rainfall in July was sparse, totaling only 3.8 cm. The most

68

significant precipitation event in 1996 was 3 consecutive days of rain from 17 - 19 June.
This rainfall may have resulted in infection of at least some gypsy moth larvae.

The low numbers of gypsy moth larvae present at Kellogg Forest may also have
contributed to the poor establishment of E. maimaiga. But Elkinton et al. (1991) found
high levels of mortality caused by E. maimaiga in plots with similarly low numbers of
gypsy moth larvae. Release plots monitored by Hajek and Roberts (1991) had successful
establishment of E. maimaiga in sites with egg mass densities as low as 600 egg
masses/ha. Egg mass densities in our release plots at Kellogg Forest were even lower,
with a maximum of 250 egg masses/ha just prior to releases of the fungus. Low larval
densities may have led to both a low rate of infection and low probabilities of finding
infected larvae during our surveys.

Resting spore counts in 1996 soil samples from Kellogg Forest were unexpected. We
expected to find resting spores in the soil collected from around the release trees, but did
not expect to find them in the other samples, especially from the control plots. Resting
spores may have been transported among the study sites on shoes between 1994 and
1996, since burlap bands were checked quite frequently. Hajek et al. (1995) showed that
this can happen, and found approximately 250 resting spores per shoe afier walking
through an area where an E. maimaiga epizootic had occurred the previous year. No
infected larvae were found in any of the plots where soil samples were tested for resting
spore density (control and E. maimaiga release plots). Therefore, we presume that the
spores Dr. Hajek found in these soil samples during the spore quantification process were
actually from the releases in 1994 and 1995. Another possibility is that the soil examined

in these tests may have contained resting spores from another species of fungus. The most

 

69

likely candidate would be Entomophaga aulicae (Reichardt in Bail) Humber, a complex of
species of which E. maimaiga is a member. The 2 species are indistinguishable without
the use of molecular techniques. E. aulicae is native to the United States and infects many
species of forest lepidoptera, but E. maimaiga is the only species of the group that is
known to infect gypsy moth (Hajek et al. 1991, Hajek et al. 1996a). However, it isn’t
likely that E. aulicae would be detected at Kellogg Forest unless an outbreak of a host had
occurred recently. The only outbreak of a hardwood defoliator at Kellogg Forest in the
last 10 yrs was orangestriped oakworm (Anisota senatoria (J. E. Smith)) in 1986 in 1 of
the English oak plantations (G. Kowalewski, Kellogg Forest Manager, pers. comm).
Therefore, it is likely that resting spores found in our soil samples were E. maimaiga
rather than E. aulicae.

It is impossible to say with certainty that the E. maimaiga infections detected at
Kellogg Foest were a result of our introduction efforts, since infected larvae were not
found within E. maimaiga release plots. We didn’t repeat the introduction in 1996, but it
may be possible that resting spores remain viable in the soil for several years. Observers
checking burlap bands in 1995 and 1996 could have inadvertently transferred resting
spores on their shoes from a release site to other locations in the forest, where other larvae
could have become infected. Airborne dispersal of conidia is another possibility (Hajek et
al. 1995). Entomophaga maimaiga epizootics were especially prevalent in Michigan in
1996, even present in central Kalamazoo County and some surrounding counties within 40
km of Kellogg Forest (Mott and Smitley, unpubl. data). Conidia may have spread within
Kellogg Forest, or to the forest from a surrounding area. Another explanation for the

spread is that early instars were infected by resting spores in a release plot, and then

7O

dispersed by wind to other locations within the forest. Other larvae could then have been
infected by conidia produced by these larvae.

The actual introductions of E. maimaiga and C. melanoscela were relatively simple
and required little effort. Permits from USDA APHIS were needed for both agents.
Direct monetary costs for the E. maimaiga introduction were very low, consisting
primarily of the cost of shipping the soil. Labor included collecting, “releasing”, and
watering the soil. The soil was also tested for the presence of plant parasitic nematodes.
The risk of introducing soil-borne pests to new areas could potentially be alleviated by
using infected gypsy moth cadavers instead of soil for E. maimaiga introductions. The
initial cost for purchase of C. melanoscela from a commercial supplier was quite high (ca.
$ 0.50 per wasp). Labor included releasing the wasps. Neither agent had a significant
effect on the gypsy moth during this study, but their long-term benefits may be more
substantial if they become established. Entomophaga maimaiga has proven its value in
causing gypsy moth population collapses across the northeastern United States and
Michigan (Elkinton et al. 1991, Hajek and Roberts 1991, Smitley et al. 1995). Inundative
releases of C. melanoscela have not been shown to have a significant impact on gypsy
moth populations based on egg mass counts, even though increased rates of parasitism
have been observed after releases of C. melanoscela (Grimble 1975, Hoy 1975, Weseloh
and Anderson 197 5, Ticehurst and Fusco 1976, Blumenthal et al. 1981, Kolodny-Hirsch et
al. 1988). The high costs of purchasing these parasitoids from commercial suppliers may
outweigh their later benefits.

Predators, parasites, and pathogens caused mortality in at least 6.5% of the gypsy

moth larvae seen under burlap bands in this study. This value is likely an underestimate,

71

since larvae that appeared healthy were not collected, and the cause of mortality for many
collected larvae could not be determined. C ampsilura concinnata provided the biggest
single source of gypsy moth larval mortality in this study. Opportunistic predators like
pentatomids were frequently observed feeding on gypsy moth larvae. NPV was present,
but likely will not cause significant gypsy moth mortality until an outbreak occurs. The
egg parasitoid Ooencyrtus kuvanae, and several parasitic tachinids specific to gypsy moth
were already present in this area. Although no single species appeared to be controlling
the gypsy moth population at Kellogg Forest, it is this combination of predators,
parasitoids, and pathogens working in concert that may limit the rate of increase of gypsy

moth population levels (Reardon 1976).

 

 

72

5:25 a: 8: I: a: :8: .882 88x8: :56 V83 3:8 :22: mm

 

.828 q: :3: 2: 3m 32 85:2 288 8: 3:8 8: a.
83852:: mm: 2: we :2 N8: 8:383 8:: 333 .8588 28 8: 8
88852:: SN 8: :2 8.8 $2 8:985 :8 8: mm

:58: n: k _ 2 a: 4. a 82 8:98: 6:5 32:8» mm

3.83.: 5.2 $2 :2 _ w.mm 32 .8282 88:88 5E9: no: imam: 509me mm
3:888:21: q: 2 : of 9mm 92 8:38: :56 6:83 35:3 :85 a

385 N: 8: G a: 32 8:955 :8 8:8: 2

3.35.: 3 $2 3 2 :3: 8:58: :8 5:8: 2

::o::::8: AEV émEofi AEoV QENEV Ewto “8:9: _::2:: 3682: E8802: ::::_::on o: 2::

3:585 52:: 8.0.59 Imm— .w>< :8: _:m:m ::::m S 55355

 

385:6 :2 :8 Wm 3:2 :: :08: z: :8 3:588 88> Ewmon ::: 53:2:
450 8:82,: .:8: _:m:m .259: E853: “EM: 2: :01: 89:28 38:82: E: 8:286:35 38m 1:: b:E:::m .m 2:: 1.1

73

Table 2. Average pheromone trap and egg mass survey counts (moths/trap and egg
masses/ha, respectively). "Forest-wide" counts include 68 traps and 33 egg mass
survey locations. "Treatment plots" only include the nine biocontrol plots. Egg
mass counts are for the year in which the eggs were laid.

 

 

 

 

Year Pheromone traps Egg mass counts
Forest-wide Treatment plots Forest-wide Treatment plots

1993 136 278 41 148

1994 72 176 52 93

1995 1 14 158 87 113

1996 338 275

 

Table 3. Total rainfall and deviation from the 50-year
average (in cm) for May and June 1994 and 1995 at the
W. K. Kellogg Experimental Forest, Michigan.

 

 

 

 

Total rainfall Deviation from average
Year May June May June
1994 3.2 14.9 -5.7 +5.2
1995 10.4 8.4 +1.5 -l.3

1996 7.6 14.4 -1.2 +4.7

 

74

 

y—a
O
LII
l
T

15 - -.- - 1995
_._.u»4

\l \O
U. 0
L l
1 I

 

 

 

 

 

4 4—
21-May lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug

 

105 ~
90 “-
75 n
60 -
45»
3o“
15—L
0 W: 1% 4-
21-May lO-Jun 3o-Jun 20-Jul 9-Aug 29-Aug

22 ' "' ' 1995
+ 1994

l

I

No. of larvae

 

 

 

 

m5»
90-
875‘
-604
"5
.45-
9
z30-
u-
0..

21-May lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug
Date

53 - -o- - 1995

T

 

 

 

 

Figure 1. Total counts of live gypsy moth larvae under burlap bands in the C.
melanoscela release plots at Kellogg Forest in 1994 and 1995. Plot identification numbers
are in the upper left corners. n = 18 bands.

 

75

 

105 4~

90 *-
75 fl + 1994

21 ' "' ’ 1995

60 ‘~
45 ~~
30 ~-
15 ~~
0 m t +4-

21-May lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug

N o. of larvae

 

 

 

 

105 *—

90 ~~
75 _L + 1994

25 ' ‘0' " 1995

60 ~~
45 -.
30 ~~
15 .4-
0 4 :o-l
21-May lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug

No. of larvae

 

 

 

 

 

 

105 4"
90 _‘__ 40 ' -.' ' 1995

75 n + 1994

60 ~-
45 ~—
30 ~-
15 «- .' ' ' .
0 ~—" ' 1 1 P +

21-May lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug
Date

No. of larvae

 

 

 

 

Figure 2. Total counts of live gypsy moth larvae under burlap bands in the E. maimaiga
release plots at Kellogg Forest in 1994 and 1995. Plot identification numbers are in the
upper left corners. n = 18 bands.

76

 

105
90 -
75 ~
60 ~
45 ~
30 ~
15 -

No. of larvae

1

 

 

0

 

21-May

 

lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug

 

105 1

90 .1-

75-
60-
45-

N o. of larvae

154

30 -~

r

T

l

I

_

 

0..

My g.

21-May

- -o- - 1995
1994

+

 

 

lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug

 

105
90 -
75 4
60 —
45 -
30 -
15 -

N o. of larvae

.—

h.-

 

0-

42

21-May

 

 

' '°' ' 1995
I 1994
l.\
o. “ 3‘
'- v' \
I ‘.~
" . . a 4
lO-Jun 30-Jun 20-Jul 9-Aug 29-Aug

Date

Figure 3. Total counts of live gypsy moth larvae under burlap bands in the control plots at
Kellogg Forest in 1994 and 1995. Plot identification numbers are in the upper left
corners. n = 18 bands.

iT‘-;1
‘C

SUMMARY

Results from our evaluation of 3 egg mass survey methods showed that egg mass
counts from the fixed-radius plots and lOO-tree plots were significantly associated with

defoliation, but timed walk counts were not. However, when we used common egg mass

 

density thresholds to evaluate the study stands for hypothetical suppression, the fixed-
radius plots and timed walks yielded similar results. The lOO-tree plot method provided
fewer overspraying errors than the other 2 methods. The time required for each method
varied substantially: the timed walk always required only 10 min, while the fixed-radius
and lOO-tree plots required 18 and 30 min, respectively. Entomophaga maimaiga and
NPV were present in most stands and likely contributed to unexpectedly low defoliation
levels.

The Sunfleck Ceptometer was a cost-efficient method of determining LAI. Since it
measures transmittance with 80 sensors at once, we collected data representing 25 ,600
point samples in each plot in an average of only 14 min. Stand LAI estimates were
relatively consistent during the period of fill leaf expansion, suggesting that the
ceptometer could be used to consistently quantify insect defoliation.

Establishment of the biocontrol agents proved to be difficult in the low-density gypsy
moth populations at Kellogg Forest. Few C otesia melanoscela were recovered, possibly

because of hyperparasitism or difficulty in successfially attacking late instars due to poor

77

78

sychronization with the gypsy moth life cycle. Difficulty in locating hosts in the low gypsy
moth densities may also have contributed to the poor establishment of C. melanoscela.

Although E. maimaiga was introduced at Kellogg Forest via resting spores in soil in
1994 and 1995, we did not find infected gypsy moth larvae until 1996. Soil used in the
1994 introduction had a very low concentration of resting spores (62.3 i 13.1 spores/g),
but the inoculum in 1995 had high numbers of resting spores (1676 d: 186 spores/g) that
were shown to be viable through a bioassay. May and June rainfall at Kellogg Forest was
highest in 1996 with 22 cm, compared to less than 19 cm in 1994 and 1995. Since
infected larvae weren’t collected in the same plots where the fungus was released, we can
not be sure that the infections were the result of our introductions. Resting spores may
have been transported on the shoes of workers, or conidia may have blown in from
surrounding areas where E. maimaiga epizootics were present.

Predators, parasites, and pathogens caused mortality to at least 6.5% of the gypsy
moth larvae seen under burlap bands in this study. Compsilura concinnata provided the
largest single source of gypsy moth larval mortality in this study. Other natural enemies
that we found included several other parasitic tachinid species, pentatomids, Ooencyrtus
kuvanae, and NPV.

The high cost of purchasing biocontrol parasitoids like C. melanoscela may not be
economically justified for low-density gypsy moth populations like those at Kellogg

Forest. E. maimaiga may be more practical, partly because of its low cost of introduction.

APPENDIX 1

APPENDIX 1

Record of Deposition of Voucher Specimens*

The specimens listed on the following sheet(s) have been deposited in
the named museum(s) as samples of those species or other taxa which were
used in this research. Voucher recognition labels bearing the Voucher
No. have been attached or included in fluid-preserved specimens.

Voucher No.: 1997'1

Title of thesis or dissertation (or other research projects):

Evaluation of Three Egg Mass Survey Methods and Two Biological
Control Agents For Gypsy Moth Management

Museum(s) where deposited and abbreviations for table on following sheets:
Entomology Museum, Michigan State University (MSU)

Other Museums:

Investigator's Name (5) (typed)

 

Lyle Jay Buss

 

Date 18 August 1997

*Reference: Yoshimoto, C. M. 1978. Voucher Specimens for Entomology in
NOrth America. Bull. Entomol. Soc. Amer. 24:141-42.

Deposit as follows:

 

Original: Include as Appendix 1 in ribbon copy of thesis or
dissertation.

Copies: Included as Appendix 1 in copies of thesis or dissertation.
Museum(s) files.
Research project files.

This form is available from and the Voucher No. is assigned by the Curator,
Michigan State University Entomology Museum.

79

1» of 1 Pages

APPZHDIX 1.1

Vbueher Specimen Data

Page

 

 

 

32 882 2 38

 

is: \$§§%

 

muunuo>aca ouuum confines: usu a: agnonuv
new nauseouaa vouowa u>oau as» vo>auoux

 

firmmmfi .oz nonu=o> Avonhuv

mmam a:: m.»:
Anyone: u.uou:w«uuu>au

Annuaaouoc «a nuance Hucouudvvu 009v

 

m m saw: an: o:
.<a .meanmwuc:4 .qaoeo
.usmz sue: awn»: .:cowp:z scam

m m m:>e:_
sues amaam seem .mmmH was: mm
.:um=m=< ..ou oo~:s:_:¥ "new:

m m m mom: .m=<-s_=s
.eoma .m=<-s_=s ..ou Engage
:c: .00 oo~:s:_:x “gov:

AmeanmNummV

:_mumoc:pms :vmmpou

Acmmwwzv :u:=cpucou :L=_wmasou

.4 e:nmw: :Pauc:s>4

 

 

 

 

 

 

 

 

co 0.. vauaooauv van vow: no 3500300
m e r r u u e .m u acusauoan pom muse Honda
e r o.d e .1 a s

 

sauna noguo no aoauonw

 

.0
‘H

o noelaz

 

 

 

80

LIST OF REFERENCES

 

LIST OF REFERENCES

Abrahamson, L. P. 1987. Five-minute walk method of sampling gypsy moth egg mass
densities. Gypsy Moth News 14.

Andreadis, T. G. and R. M. Weseloh. 1990. Discovery of Entomophaga maimaiga in
North American gypsy moth, Lymantria dispar. Proc. Natl. Acad. Sci. USA 87:
2461-2465.

Barbosa, P., J. L. Capinera and E. A. Harrington. 197 S. The gypsy moth parasitoid
complex in western Massachusetts: a study of parasitoids in areas of high and low
host density. Environ. Entomol. 4: 842-846.

Bell, R A., C. D. Owens, M. Shapiro and J. G. R. Tardif. 1981. Development of mass
rearing technology. In C. C. Doane and M. L. McManus [eds], The Gypsy Moth:
Research Toward Integrated Pest Management. US. Dept. Agn'c. Tech. Bull. 1584:
599-633.

Blumenthal, E. M., R. A. Fusco and R. C. Reardon. 1981. Parasite augmentation. In C.
C. Doane and M. L. McManus [eds], The Gypsy Moth: Research Toward Integrated
Pest Management. US. Dept. Agric. Tech. Bull. 1584: 402-408.

Bolstad, P. V. and S. T. Gower. 1990. Estimation of leaf area index in fourteen southern
Wisconsin forest stands using a portable radiometer. Tree Physiol. 7: 115-124.

Burton, A. J ., K. S. Pregitzer and D. D. Reed. 1991. Leaf area and foliar biomass
relationships in northern hardwood forests located along an 800 km acid deposition
gradient. For. Sci. 37: 1041-1059.

Butalla, H. A. 1996. The GM Digest. <http://www.fsl.wvnet.edu/fiip/gm_digest.html>.
Accessed 1 March 1997.

Cameron, E. A. and R. M. Reeves. 1990. Carabidae (Coleoptera) associated with gypsy
moth, Lymantria dispar (L.) (Lepidoptera: Lymantriidae), populations subjected to
Bacillus thuringiensis Berliner treatments in Pennsylvania. Can. Ent. 122: 123-129.

81

82

Carter, M. R., F. W. Ravlin and M. L. McManus. 1991. Changes in gypsy moth
(Lepidoptera: Lymantriidae) fecundity and male wing length resulting from defoliation.
Environ. Entomol. 20: 1042-1047.

Chason, J. W., D. D. Baldocchi and M. A. Huston. 1991. A comparison of direct and
indirect methods for estimating forest canopy leaf area. Agric. For. Meteorol. 57:
107-128.

Chilcote, C. A., M. Scriber and G. Fowler. 1994. Development of an efficient gypsy
moth sampling protocol for forests and suburban areas. P. 13-14 in S. L. C. Fosbroke
and K. W. Gottschalk, eds. Proceedings, US. Department of Agriculture Interagency
Gypsy Moth Research Forum 1994. USDA Gen. Tech. Rep. NEE-188. m,

Clinton, B. D. 1995. Temporal variation in photosynthetically active radiation (PAR) in
mesic southern Appalachian hardwood forests with and without Rhododendron
understories. pp. 534-540 In: Proceedings 10th Central Hardwood Forest
Conference, March 5-8, 1995, Morgantown, West Virginia. NE For. Exp. Sta. Gen.
Tech. Rep. NE-197.

.1 ~ - e
.-
‘-

Dalla-Tea, F. and E. J. Jokela. 1991. Needlefall, canopy light interception, and
productivity of young intensively managed slash and loblolly pine stands. For. Sci. 37:
1298-1313.

Decagon Devices, Inc. 1987. Sunfleck PAR Ceptometer Plant Canopy Measurement
Operator’s Manual. Pullman, Washington USA.

Eggen, D. A. and L. P. Abrahamson. 1983. Estimating gypsy moth egg mass densities.
School of Forestry Miscellaneous Publication (ESF 83-002), State University of New
York, College of Environmental Science and Forestry, Syracuse, New York.

Elkinton, J. S., A. E. Hajek, G. H. Boettner and E. E. Simons. 1991. Distribution and
apparent spread of Entomophaga maimaiga (Zygomycetes: Entomophthorales) in

gypsy moth (Lepidoptera: Lymantriidae) populations in North America. Environ.
Entomol. 20: 1601-1605.

Fassnacht, K. S., S. T. Gower, J. M. Norman and R. E. McMurtrie. 1994. A comparison
of optical and direct methods for estimating foliage surface area index in forests.
Agn'c. For. Meteorol. 71: 183-207.

Fleischer, S. J ., F. W. Ravlin and R. C. Reardon. 1991. Implementation of sequential
sampling plans for gypsy moth (Lepidoptera: Lymantriidae) egg masses in eastern
hardwood forests. J. Econ. Entomol. 84: 1100-1107.

83

Gazarini, L. C., M. C. C. Araujo, N. Borralho and J. S. Pereira. 1990. Plant area index in
Eucalyptus globulus plantations determined indirectly by a light interception method.
Tree Physiol. 7: 107-113.

Grace, J. R. 1986. The influence of gypsy moth on the composition and nutrient content
of litter fall in a Pennsylvania oak forest. Forest Sci. 32: 855-870.

Gresham, C. A. 1982. Litterfall patterns in mature loblolly and longleaf pine stands in
coastal South Carolina. For. Sci. 28: 223-231.

Grimble, D. G. 1975. Releases of Apanteles melanoscelus for gypsy moth suppression.
SUNY College of Environ. Sci. and For. AFRI Research Note 15: 6 pp. h:

Grimble, D. G. and C. E. Palm. 1976. Mortality in the overwintering generations of
Apanteles melanoscelus. SUNY College of Environ. Sci. and For. AF RI Research
Note 21: 4 pp.

 

Hajek, A. E. and D. W. Roberts. 1991. Pathogen reservoirs as a biological control
resource: Introduction of Entomophaga maimaiga to North American gypsy moth,
Lymantria dispar, populations. Biol. Contr. 1: 29-34.

Hajek, A. E. and D. W. Roberts. 1992. Field diagnosis of gypsy moth (Lepidoptera:
Lymantriidae) larval mortality caused by Entomophaga maimaiga and the gypsy moth
nuclear polyhedrosis virus. Environ. Entomol. 21: 706-713.

Hajek, A. E. and M. M. Wheeler. 1994. Application of techniques for quantification of
soil-home Entomophthoralean resting spores. J. Invertebr. Pathol. 64: 71-73.

Hajek, A. E., R. A. Humber, J. S. Elkinton, B. May, S. R. A. Walsh and J. C. Silver.
1990a. Allozyme and restriction fragment length polymorphism analyses confirm
Entomophaga maimaiga responsible for 1989 epizootics in North American gypsy
moth populations. Proc. Natl. Acad. Sci. 87: 6979-6982.

Hajek, A. E., R. I. Carruthers and R. S. Soper. 1990b. Temperature and moisture
relations of sporulation and germination by Entomophaga maimaiga (Zygomycetes:
Entomophthoraceae), a fiingal pathogen of Lymantria dispar (Lepidoptera:
Lymantriidae). Environ. Entomol. 19: 85-90.

Hajek, A. E., R. A. Humber, S. R. A. Walsh and J. C. Silver. 1991. Sympatric
occurrence of two Entomophaga aulicae (Zygomycetes: Entomophthorales) complex
species attacking forest Lepidoptera. J. Invertebr. Pathol. 58: 373-3 80.

Hajek, A. E., T. S. Larkin, R. I. Carruthers and R. S. Soper. 1993. Modeling the
dynamics of Entomophaga maimaiga (Zygomycetes: Entomophthorales) epizootics in

84

gypsy moth (Lepidoptera: Lymantriidae) populations. Environ. Entomol. 22: l 172-
1187.

Hajek, A. E., R. A. Humber and J. S. Elkinton. 1995. Mysterious origin of Entomophaga
maimaiga in North America. Amer. Entomol. 41: 31-42.

Hajek, A. E, S. R. A. Walsh, D. R. Strong and J. C. Silver. 1996a. A disjunct
Californian strain of Entomophaga aulicae infecting Orgyia vetusta. J. Invertebr.
Pathol. 68: 260-268.

Hajek, A. E., J. S. Elkinton and J. J. Witcosky. 1996b. Introduction and spread of the
fiJngal pathogen Entomophaga maimaiga (Zygomycetes: Entomophthorales) along the :-

leading edge of gypsy moth (Lepidoptera: Lymantriidae) spread. Environ. Entomol.
25: 1235-1247.

. .

Herrick, O. W. and D. A. Gansner. 1987. Gypsy moth on a new frontier: forest tree
defoliation and mortality. North. J. Appl. For. 4: 128-133.

Hoy, M. A. 1975. Forest and laboratory evaluations of hybridized Apanteles
melanoscelus [Hym.: Braconidae], a parasitoid of Porthetria dispar [Lep.:
Lymantriidae]. Entomophaga 20: 261-268.

Hoy, M. A. 1976. Establishment of gypsy moth parasitoids in North America: an
evaluation of possible reasons for establishment or non-establishment. In J. F.
Anderson and H. K. Kaya [eds], Perspectives in Forest Entomology. New York,
Academic Press.

Huston, M. and J. G. Isebrands. 1995. Monitoring interception of photosynthetically
active radiation in forests. Bull. Ecol. Soc. Amer. 76: 95-97.

Jobidon, R. 1992. Measurement of light transmission in young conifer plantations: a new
technique for assessing herbicide efficacy. Northern J. Appl. For. 9: 112-115.

Jobin, L., C. Hebert and J .-P. Bourassa. 1992. Head capsule width of larval populations
of the gypsy moth (Lepidoptera: Lymantriidae) in Quebec, with reference to Dyar's
hypothesis. Environ. Entomol. 21: 89-93.

Kolodny-Hirsch, D. M. 1986. Evaluation of methods for sampling gypsy moth
(Lepidoptera: Lymantriidae) egg mass populations and development of sequential
sampling plans. Environ. Entomol. 15: 122-127.

Kolodny-Hirsch, D. M., R. C. Reardon, K. W. Thorpe and M. J. Raupp. 1988.
Evaluating the impact of sequential releases of Cotesia melanoscela (Hymenoptera:
Braconidae) on Lymantria diqur (Lepidoptera: Lymantriidae). Environ. Entomol.
17: 403-408.

85

Kulman, H. M. 1971. Effects of insect defoliation on growth and mortality of trees.
Ann. Rev. Entomol. 16: 289-324.

Leonard, D. E. 1981. Bioecology of the gypsy moth, pp. 9-29 In C. C. Doane and M. L.
McManus [eds], The gypsy moth: research toward integrated pest management.
USDA Forest Service Technical Bulletin 1584.

Liebhold, A. M., J. S. Elkinton, D. R. Miller and Y. S. Wang. 1988. Estimating oak leaf
area index and gypsy moth, Lymantria dispar (L.) (Lepidoptera: Lymantriidae),
defoliation using canopy photographs. Environ. Entomol. 17: 560-566.

Liebhold, A., D. Twardus and J. Buonaccorsi. 1991. Evaluation of the timed-walk
method of estimating gypsy moth (Lepidoptera: Lymantriidae) egg mass densities. J.
Econ. Entomol. 84: 1774-1781.

Liebhold, A., K. Thorpe, J. Ghent and D. B. Lyons. 1994. Gypsy moth egg mass
sampling for decision-making: a users' guide. USDA Forest Service, Forest Health
Protection, NA-TP-04-94.

Liebhold, A. M., K. W. Gottschalk, R.-M. Muzika, M. E. Montgomery, R. Young, K.
O'Day and B. Kelley. 1995. Suitability of North American tree species to the gypsy

moth: A summary of field and laboratory tests. USDA Forest Service, NE Forest
Exp. Sta. Gen. Tech. Rep. NE-211.

Maass, J. M., J. M. Vose, W. T. Swank and A. Martinez-Yrizar. 1995. Seasonal changes
of leaf area index (LAI) in a tropical deciduous forest in west Mexico. For. Ecol. and
Manage. 74: 171-180.

Mattson, W. J. and N. D. Addy. 1975. Phytophagous insects as regulators of forest
primary production. Science 190: 515-522.

McLeod, S. D. and S. W. Running. 1988. Comparing site quality indices and
productivity in ponderosa pine stands of western Montana. Can. J. For. Res. 18: 346-
352.

McPherson, J. E. 1982. The Pentatomoidea (Hemiptera) of northeastern North America
with emphasis on the fauna of Illinois. Southern Illinois Univ. Press, Carbondale, IL.
240 pp.

Onken, A., D. Twardus and D. Allen-Reid. 1996. Treatment objectives and priorities for

cooperative, other federal, and national forest gypsy moth suppression projects.
Gypsy Moth News 41: 5-15.

86

Péch, G. 1986. Mobile sampling of solar radiation under conifers. Agric. For. Meteorol.
37: 15-28.

Pierce, L. L. and S. W. Running. 1988. Rapid estimation of coniferous forest leaf area
index using a portable integrating radiometer. Ecology 69: 1762-1767.

Poinar, G. 0., Jr. and G. M. Thomas. 1978. Diagnostic Manual for the Identification of
Insect Pathogens. Plenum Press, New York. 218 pp.

Ravlin, F. W., R. G. Bellinger and E. A. Roberts. 1987. Gypsy moth management
programs in the United States: status, evaluation, and recommendations. Bulletin of
the BSA 33: 90-98.

Reardon, R. C. 1976. Parasite incidence and ecological relationships in field populations
of gypsy moth larvae and pupae. Environ. Entomol. 5: 981-987.

Sabrosky, C. W. and R. C. Reardon. 1976. Tachinid parasites of the gypsy moth,
Lymantria dispar, with keys to adults and puparia. Misc. Pub. Entomol. Soc. Am.
10(2): 126 pp.

SAS Institute. 1995. JMP statistics and graphics guide. SAS Institute Inc., Cary, NC.

Skaller, P. M. 1985. Patterns in the distribution of gypsy moth (Lymantria dispar)
(Lepidoptera: Lymantriidae) egg masses over an 11-year population cycle. Environ.
Entomol. 14: 106-117.

Smith, H. R. and R. A. Lautenschlager. 1978. Predators of the gypsy moth. Agric.
Handbook 534. 72 pp.

Smitley, D. R., L. S. Bauer, A. E. Hajek, F. J. Sapio and R. A. Humber. 1995.
Introduction and establishment of Entomophaga maimaiga, a fiingal pathogen of
gypsy moth (Lepidoptera: Lymantriidae) in Michigan. Environ. Entomol. 24: 1685-
1695.

Sokal, R. R and F. J. Rohlf. 1995. Biometry. 3rd edition. W. H. Freeman and
Company, New York. 887 pp.

Swank, W. T., J. B. Waide, D. A. Crossley, Jr. and R. L. Todd. 1981. Insect defoliation
enhances nitrate export from forest ecosystems. Oecologia 51: 297-299.

Talerico, R. L. 1981. Defoliation as an indirect means of population assessment. In C. C.
Doane and M. L. McManus [eds], The Gypsy Moth: Research Toward Integrated
Pest Management. US. Dept. Agric. Tech. Bull. 1584: 38-49.

87

Tallent-Halsell, N. G. (ed). 1994. Forest Health Monitoring 1994 Field Methods Guide.
EPA/620/R-94/027. US. Environmental Protection Agency, Washington, DC. 277

PP-

Ticehurst, M. and R. A. Fusco. 1976. Release of three Apanteles spp. against the gypsy
moth in Pennsylvania. J. Econ. Entomol. 69: 307-308.

Van Sickle, D. and R. M. Weseloh. 1974. Habitat variables that influence the attack by
hyperparasites of Apanteles melanoscelus cocoons. J. N. Y. Entomol. Soc. 82: 2-5.

Vose, J. M. and W. T. Swank. 1990. Assessing seasonal leaf area dynamics and vertical
leaf area distribution in eastern white pine (Pinus strobus L.) with a portable light
meter. Tree Physiol. 7: 125-134.

Vose, J. M., N. H. Sullivan, B. D. Clinton and P. W. Bolstad. 1995. Vertical leaf area
distribution, light transmittance, and application of the Beer-Lambert Law in four

mature hardwood stands in the southern Appalachians. Can. J. For. Res. 25: 1036-
1043.

Wallner, W. E. and M. L. McManus. 1981. Summary. In C. C. Doane and M. L.
McManus [eds], The Gypsy Moth: Research Toward Integrated Pest Management.
US. Dept. Agric. Tech. Bull. 1584: 202-203.

Wargo, P. M. 1977. Armillariella mellea and Agrilus bilineatus and mortality of
defoliated oak trees. For. Sci. 23: 485-492.

Weseloh, R. M. 1976. Reduced effectiveness of the gypsy moth parasite, Apanteles
melanoscelus, in Connecticut due to poor seasonal synchronization with its host.
Environ. Entomol. 5: 743-746.

Weseloh, R. M. 1981. Apanteles melanoscelus (Ratzeburg). In C. C. Doane and M. L.
McManus [eds], The Gypsy Moth: Research Toward Integrated Pest Management.
US. Dept. Agric. Tech. Bull. 1584: 399-402.

Weseloh, R. M. 1983. Population sampling method for cocoons of the gypsy moth
(Lepidoptera: Lymantriidae) parasite, Apanteles melanoscelus (Hymenoptera:
Braconidae), and relationship of its population levels to predator- and hyperparasite-
induced mortality. Environ. Entomol. 12: 1228-1231.

Weseloh, R. M. and J. F. Anderson. 1975. Inundative release of Apanteles melanoscelus
against the gypsy moth. Environ. Entomol. 4: 33-36.

Williams, D. W., R. W. Fuester, W. W. Metterhouse, R. J. Balaam, R. H. Bullock and R.
J. Chianese. 1991. Oak defoliation and population density relationships for the gypsy
moth (Lepidoptera: Lymantriidae). J. Econ. Entomol. 84: 1508-1514.

 

88

Wilson, R. W., Jr. and G. A. Fontaine. 1978. Gypsy moth egg-mass sampling with fixed-
and variable-radius plots. Agric. Handbook No. 523. 47 pp.

"Illlllllllllllllllllls